1 2017-11-15 Jakub Jelinek <jakub@redhat.com>
4 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6 (can_widen_mult_without_libcall): New function.
7 (expand_mul_overflow): If only checking unsigned mul overflow,
8 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
9 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
10 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
11 (expand_DIVMOD): Formatting fix.
12 * expmed.h (expand_mult): Add NO_LIBCALL argument.
13 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
14 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
16 PR tree-optimization/82977
17 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
18 constructed temporary to strlen_to_stridx.put.
20 2017-11-15 Martin Liska <mliska@suse.cz>
22 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
23 * configure: Regenerate.
25 2017-11-15 Martin Liska <mliska@suse.cz>
28 * config/sh/sh-mem.cc: Use proper probability for
31 2017-11-14 Jeff Law <law@redhat.com>
33 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
34 the red zone for stack_clash_protection_final_dynamic_probe targets
35 when the total dynamic stack size is zero bytes.
37 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
40 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
42 * dumpfile.h (TDF_COMPARE_DEBUG): New.
43 * final.c (rest_of_clean_state): Set it for the
45 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
46 class when TDF_COMPARE_DEBUG is set.
48 * dwarf2out.c (gen_producer_string): Discard
51 2017-11-15 Joseph Myers <joseph@codesourcery.com>
54 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
55 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
56 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
57 (__TGMATH_REAL_2_3): Remove macros.
58 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
59 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
61 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
63 (remquo): Define using __TGMATH_REAL_3.
65 2017-11-14 Jeff Law <law@redhat.com>
67 * vr-values.c: New file with contents extracted from tree-vrp.c.
68 * Makefile.in (OBJS): Add vr-values.o
69 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
70 (set_value_range, set_and_canonicalize_value_range): Likewise.
71 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
72 (value_range_constant_singleton, symbolic_range_p): Likewise.
73 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
74 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
75 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
76 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
77 (find_case_label_range, find_case_label_index): Likewise.
78 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
79 (range_int_cst_singleton_p, value_inside_range): Likewise.
80 (get_single_symbol): Likewise.
81 (switch_update): Move structure definition here.
82 (to_remove_edges, to_update_switch_stmts): Provide externs.
83 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
84 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
85 (vrp_val_is_min, set_value_range): Likewise.
86 (set_and_canonicalize_value_range, copy_value_range): Likewise.
87 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
88 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
89 (range_is_nonnull, range_int_cst_p): Likewwise.
90 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
91 (get_single_symbol, operand_less_p): Likewise
92 (compare_values_warnv, compare_values): Likewise.
93 (value_inside_range, value_range_constant_singleton): Likewise.
94 (zero_nonzero_bitgs_from_vr): Likewise.
95 (extract_range_from_binary_expr_1): Likewise.
96 (overflow_comparison_p): Likewise.
97 (to_remove_edges, to_update_switch_stmts): Likewise.
98 (find_case_label-index, find_case_label_range): Likewise.
99 (switch_update, set_value_range_to_nonnegative): Remove.
100 (set_value_range_to_truthvalue): Likewise.
101 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
102 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
103 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
104 (find_case_label_ranges, test_for_singularity): Likewise.
105 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
106 (x_vr_values): Move to its remaining use site.
108 2017-11-10 Jeff Law <law@redhat.com>
110 * vr-values.h (VR_INITIALIZER): Move #define here.
111 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
112 * Makefile.in (OBJS): Add tree-evrp.o
113 * tree-vrp.h (assert_info): Move structure definition here.
114 (set_value_range_to_varying): Prototype.
115 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
116 (infer_value_range, register_edge_assert_for): Likewise.
117 (stmt_interesting_for_vrp): Likewise.
118 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
119 (set_value_range_to_varying): No longer static.
120 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
121 (infer_value_range, register_edge_assert_for): Likewise.
123 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
125 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
126 generate the XXBRD instruction.
128 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
129 (rs6000_builtin_type_compatible): Treat _Float128 and long double
130 as being compatible if -mabi=ieeelongdouble.
131 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
132 to setup float128 built-ins with hardware support.
133 (BU_FLOAT128_HW_2): Likewise.
134 (BU_FLOAT128_HW_3): Likewise.
135 (BU_FLOAT128_HW_VSX_1): Likewise.
136 (BU_FLOAT128_HW_VSX_2): Likewise.
137 (scalar_extract_expq): Change float128 built-in functions to
138 accommodate having both KFmode and TFmode functions. Use the
139 KFmode variant as the default.
140 (scalar_extract_sigq): Likewise.
141 (scalar_test_neg_qp): Likewise.
142 (scalar_insert_exp_q): Likewise.
143 (scalar_insert_exp_qp): Likewise.
144 (scalar_test_data_class_qp): Likewise.
145 (sqrtf128_round_to_odd): Delete processing the round to odd
146 built-in functions as special built-in functions, and define them
147 as float128 built-ins. Use the KFmode variant as the default.
148 (truncf128_round_to_odd): Likewise.
149 (addf128_round_to_odd): Likewise.
150 (subf128_round_to_odd): Likewise.
151 (mulf128_round_to_odd): Likewise.
152 (divf128_round_to_odd): Likewise.
153 (fmaf128_round_to_odd): Likewise.
154 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
155 support for KFmode and TFmode xststdcqp calls.
156 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
157 point, switch the built-in handlers for the get/set float128
158 exponent, get float128 mantissa, float128 test built-ins, and the
159 float128 round to odd built-in functions. Eliminate creating the
160 float128 round to odd built-in functions as special built-ins.
161 (rs6000_init_builtins): Eliminate special creation of the float128
162 round to odd built-in functions.
163 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
164 function insns to support both TFmode and KFmode variants.
165 (xsxsigqp_<mode>): Likewise.
166 (xsiexpqpf_<mode>): Likewise.
167 (xsiexpqp_<mode>): Likewise.
168 (xststdcqp_<mode>): Likewise.
169 (xststdcnegqp_<mode>): Likewise.
170 (xststdcqp_<mode>): Likewise.
172 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
174 * tree-ssa-threadupdate.c (compute_path_counts): Remove
175 unused path_in_freq_ptr parameter.
176 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
178 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
180 * ipa-inline.c (edge_badness): Dump sreal frequency.
181 (compute_inlined_call_time): Match natural implementaiton ...
182 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
183 forgotten division by CGRAPH_FREQ_BASE.
185 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
187 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
188 Solaris 11. Update comment.
189 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
191 * config/sol2.h (STARTFILE_SPEC): Likewise.
192 * configure: Regenerate.
194 2017-11-14 Carl Love <cel@us.ibm.com>
196 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
197 le_ and be_ prefixes to swap* variables. Remove
198 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
201 2017-11-14 Jason Merrill <jason@redhat.com>
203 Support GTY((cache)) on hash_map.
204 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
205 (ggc_cache_remove): Override it instead of ggc_mx.
206 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
207 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
208 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
209 (simple_cache_map_traits): Override maybe_mx.
210 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
211 (hash_map): Friend gt_cleare_cache.
212 (gt_cleare_cache): New.
213 * tree.h (tree_cache_traits): New hash_map traits class.
214 (tree_cache_map): New typedef.
216 2017-11-14 Richard Biener <rguenther@suse.de>
218 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
219 paramter and handling.
220 (cleanup_control_flow_bb): Likewise.
221 (cleanup_control_flow_pre): New helper performing a DFS walk
222 to call cleanup_control_flow_bb in PRE order.
223 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
224 via cleanup_control_flow_pre.
226 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
228 * config/aarch64/aarch64-simd.md
229 (aarch64_simd_bsl<mode>_internal): Remove DImode.
230 (*aarch64_simd_bsl<mode>_alt): Likewise.
231 (aarch64_simd_bsldi_internal): New.
232 (aarch64_simd_bsldi_alt): Likewise.
234 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
236 * tracer.c (better_p): Do not compare frequencies.
237 * reg-stack.c (better_edge): Likewise.
238 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
241 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
243 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
244 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
245 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
246 * lto-streamer-in.c (input_function): Use update_max_bb_count.
247 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
248 * predict.c (maybe_hot_frequency_p): Inline to ...
249 (maybe_hot_count_p): ... here; rewrite to counts.
250 (counts_to_freqs): Rename to ...
251 (update_max_bb_count): ... this one.
252 (expensive_function_p): Use counts.
253 (estimate_bb_frequencies): Update.
254 (rebuild_frequencies): Update.
255 * predict.h (counts_to_freqs): Rename to ...
256 (update_max_bb_count): ... this one.
257 * profile.c (compute_branch_probabilities): Add debug info
258 * tree-inline.c (expand_call_inline): Update debug info.
259 (optimize_inline_calls): Use update_max_bb_count..
260 (tree_function_versioning): Use update_max_bb_count..
261 * value-prof.c (gimple_value_profile_transformations):
262 Do not use update_max_bb_count.
264 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
266 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
267 always use frequencies.
269 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
271 * bb-reorder.c: Remove frequencies from comments.
272 (better_edge_p): Use profile counts.
273 (find_traces): Dump profile counts.
274 (rotate_loop): Use profile counts.
275 (find_traces_1_round): Likewise.
276 (connect_better_edge_p): Use counts instead of probabilities for
278 (copy_bb_p): Drop early check for non-0 frequency.
279 (sanitize_hot_paths): Update comments.
281 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
283 * ipa-split.c (struct split_point): Add count.
284 (consider_split): Do not compute incoming frequency; compute incoming
285 count and store it to split_point.
286 (split_function): Set count of the call to split part correctly.
288 2017-11-13 Carl Love <cel@us.ibm.com>
290 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
292 2017-11-13 Tom Tromey <tom@tromey.com>
294 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
296 2017-11-13 Carl Love <cel@us.ibm.com>
298 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
299 Add support for builtins:
300 unsigned int vec_first_{,miss}_match_{,or_eos}index,
301 vector {un,}signed {char,int,short},
302 vector {un,}signed {char,int,short}) arguments.
303 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
304 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
305 Add BU_P9V_AV_2 expansions for the builtins.
306 * config/rs6000/altivec.h (vec_first_match_index,
307 vec_first_mismatch_index, vec_first_match_or_eos_index,
308 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
309 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
310 new extern declaration.
311 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
312 * config/rs6000/vsx.md (first_match_index_<mode>,
313 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
314 first_mismatch_or_eos_index_<mode>): Add define expand.
315 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
316 * doc/extend.texi: Update the built-in documenation file for the new
319 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
321 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
322 into the min/max operations for _Float<N> and _Float<N>X types.
324 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
327 * dwarf2out.c (do_eh_frame): New static variable.
328 (dwarf2out_begin_prologue): Set it.
329 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
331 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
333 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
335 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
337 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
340 * debug.h (dwarf2out_do_eh_frame): Declare.
341 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
342 (dwarf2out_do_frame): Use it.
343 (dwarf2out_do_cfi_asm): Likewise.
344 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
345 (dwarf2out_assembly_start): Likewise.
346 (dwarf2out_begin_prologue): Fix comment.
347 * toplev.c (compile_file): Always call dwarf2out_frame_finish
348 if the target needs either debug or unwind DWARF2 info.
349 * lto-opts.c (lto_write_options): Do not save -fexceptions,
350 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
351 -ftrapping-math, -ftrapv and -fwrapv.
353 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
355 * cgraph.c (cgraph_edge::sreal_frequency): New function.
356 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
357 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
358 (estimate_edge_size_and_time): Likewise.
359 (ipa_merge_fn_summary_after_inlining): Likewise.
360 * ipa-inline.c (cgraph_freq_base_rec): Remove.
361 (compute_uninlined_call_time): Use sreal_frequency.
362 (compute_inlined_call_time): Likewise.
363 (ipa_inline): Do not initialize cgraph_freq_base_rec.
364 * profile-count.c: Include sreal.h.
365 (profile_count::to_sreal_scale): New.
366 * profile-count.h: Forward declare sreal.
367 (profile_count::to_sreal_scale): Declare.
369 2017-11-13 Nathan Sidwell <nathan@acm.org>
371 * diagnostic.c (maybe_line_and_column): New.
372 (diagnostic_get_location_text): Use it.
373 (diagnostic_report_current_module): Likewise.
374 (test_diagnostic_get_location_text): Add tests.
376 2017-11-13 Luis Machado <luis.machado@linaro.org>
378 * doc/md.texi (Specifying processor pipeline description): Fix
379 incorrect latency for the div instruction example.
381 2017-11-13 Jakub Jelinek <jakub@redhat.com>
383 PR tree-optimization/78821
384 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
385 that bit_not_p is the same.
386 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
387 (split_group): Count precisely bit_not_p bits in each statement.
388 (invert_op): New function.
389 (imm_store_chain_info::output_merged_store): Use invert_op to
390 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
391 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
393 2017-11-13 Martin Liska <mliska@suse.cz>
395 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
396 (struct source_info): Likewise.
397 (add_branch_counts): Likewise.
398 (add_line_counts): Likewise.
399 (function_summary): Likewise.
400 (output_intermediate_line): Likewise.
401 (generate_results): Likewise.
403 2017-11-13 Martin Liska <mliska@suse.cz>
405 * gcov.c (struct block_info): Remove typedef for block_t.
406 (struct line_info): Likewise.
407 (line_info::has_block): Likewise.
408 (EXIT_BLOCK): Likewise.
411 (get_cycles_count): Likewise.
412 (process_file): Likewise.
413 (read_graph_file): Likewise.
414 (solve_flow_graph): Likewise.
415 (find_exception_blocks): Likewise.
416 (add_line_counts): Likewise.
417 (accumulate_line_info): Likewise.
418 (output_line_details): Likewise.
420 2017-11-13 Martin Liska <mliska@suse.cz>
422 * gcov.c (struct arc_info): Remove typedef for arc_t.
423 (struct line_info): Likewise.
424 (add_branch_counts): Likewise.
425 (output_branch_count): Likewise.
426 (function_info::~function_info): Likewise.
428 (output_intermediate_line): Likewise.
429 (read_graph_file): Likewise.
430 (solve_flow_graph): Likewise.
431 (find_exception_blocks): Likewise.
432 (add_line_counts): Likewise.
433 (accumulate_line_info): Likewise.
434 (output_line_details): Likewise.
435 (output_function_details): Likewise.
437 2017-11-13 Martin Liska <mliska@suse.cz>
439 * gcov.c (struct function_info): Remove typedef for function_t.
440 (struct source_info): Likewise.
441 (source_info::get_functions_at_location): Likewise.
442 (solve_flow_graph): Likewise.
443 (find_exception_blocks): Likewise.
444 (add_line_counts): Likewise.
445 (output_intermediate_file): Likewise.
446 (process_file): Likewise.
447 (generate_results): Likewise.
448 (release_structures): Likewise.
449 (read_graph_file): Likewise.
450 (read_count_file): Likewise.
451 (accumulate_line_counts): Likewise.
452 (output_lines): Likewise.
454 2017-11-13 Martin Liska <mliska@suse.cz>
456 * gcov.c (function_info::function_info): Remove num_counts
457 and add vector<gcov_type>.
458 (function_info::~function_info): Use the vector.
459 (process_file): Likewise.
460 (read_graph_file): Likewise.
461 (read_count_file): Likewise.
462 (solve_flow_graph): Likewise.
464 2017-11-13 Martin Liska <mliska@suse.cz>
466 * gcov.c (function_info::is_artificial): New function.
467 (process_file): Erase all artificial early.
468 (generate_results): Skip as all artificial are already
471 2017-11-13 Martin Liska <mliska@suse.cz>
473 * gcov.c (read_graph_file): Store to global vector of functions.
474 (read_count_file): Iterate the vector.
475 (process_file): Likewise.
476 (generate_results): Likewise.
477 (release_structures): Likewise.
479 2017-11-13 Jakub Jelinek <jakub@redhat.com>
481 PR tree-optimization/82954
482 * gimple-ssa-store-merging.c
483 (imm_store_chain_info::coalesce_immediate_stores): If
484 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
486 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
488 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
489 Upddate call to ENDIAN_LANE_N.
490 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
491 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
492 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
493 and use aarch64_endian_lane_rtx.
494 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
496 2017-11-12 Tom de Vries <tom@codesourcery.com>
498 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
500 2017-11-12 Tom de Vries <tom@codesourcery.com>
502 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
504 2017-11-12 Tom de Vries <tom@codesourcery.com>
506 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
508 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
509 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
510 * defaults.h (ASM_OUTPUT_LABELREF): Same.
512 2017-11-11 Martin Sebor <msebor@redhat.com>
515 * doc/extend.texi (attribute nonstring): Remove spurious argument.
518 * prefic.c (translate_name): Replace strncpy with memcpy to
519 avoid -Wstringop-truncation.
521 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
523 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
525 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
527 * predict.c (maybe_hot_frequency_p): Do not use cfun.
529 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
531 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
534 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
536 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
537 enable generating XXBRH if the value is in a vector register.
538 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
539 value is in a vector register.
540 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
541 register to register bswap64's instead of doing the GPR sequence
542 used on previous machines.
543 (bswapdi2_xxbrd): New insn.
544 (bswapdi2_reg): Disallow on ISA 3.0.
545 (register to register bswap64 splitter): Do not split the insn on
546 ISA 3.0 systems that use XXBRD.
548 2017-11-10 Martin Sebor <msebor@redhat.com>
551 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
553 (find_subframework_file): Same.
555 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
557 * auto-profile.c (afdo_indirect_call): Drop frequency.
558 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
559 (cgraph_node::create_edge): Drop frequency argument.
560 (cgraph_node::create_indirect_edge): Drop frequency argument.
561 (cgraph_edge::make_speculative): Drop frequency arguments.
562 (cgraph_edge::resolve_speculation): Do not update frequencies
563 (cgraph_edge::dump_edge_flags): Do not dump frequency.
564 (cgraph_node::dump): Check consistency in IPA mode.
565 (cgraph_edge::maybe_hot_p): Use IPA counter.
566 (cgraph_edge::verify_count_and_frequency): Rename to ...
567 (cgraph_edge::verify_count): ... this one; drop frequency checking.
568 (cgraph_node::verify_node): Update.
569 * cgraph.h (struct cgraph_edge): Drop frequency.
570 (cgraph_edge::frequency): New function.
571 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
573 (cgraph_edge::rebuild_edges): Likewise.
574 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
575 (duplicate_thunk_for_node): Do not pass frequency.
576 (cgraph_node::create_clone): Scale only counts.
577 (cgraph_node::create_virtual_clone): Do not pass frequency.
578 (cgraph_node::create_edge_including_clones): Do not pass frequency.
579 (cgraph_node::create_version_clone): Do not pass frequency.
580 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
581 (cgraph_node::expand_thunk): Do not pass frequency.
582 (cgraph_node::create_wrapper): Do not pass frequency.
583 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
585 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
586 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
587 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
588 (gather_caller_stats): Use frequency function.
589 (ipcp_cloning_candidate_p): Use frequency function.
590 (ipcp_propagate_stage): Use frequency function.
591 (get_info_about_necessary_edges): Use frequency function.
592 (update_profiling_info): Update only IPA profile.
593 (update_specialized_profile): Use frequency functoin.
594 (perhaps_add_new_callers): Update only IPA profile.
595 * ipa-devirt.c (ipa_devirt): Use IPA profile.
596 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
597 (dump_ipa_call_summary): Use frequency function.
598 (estimate_edge_size_and_time): Use frequency function.
599 (ipa_merge_fn_summary_after_inlining): Use frequency function.
600 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
601 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
602 (update_noncloned_counts): ... ths one; scale counts only.
603 (clone_inlined_nodes): Do not scale frequency.
604 (inline_call): Do not pass frequency.
605 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
606 (compute_inlined_call_time): Use IPA profile.
607 (want_inline_small_function_p): Use IPA profile.
608 (want_inline_self_recursive_call_p): Use IPA profile.
609 (edge_badness): Use IPA profile.
610 (lookup_recursive_calls): Use IPA profile.
611 (recursive_inlining): Do not pass frequency.
612 (resolve_noninline_speculation): Do not update frequency.
613 (inline_small_functions): Collect max of IPA profile.
614 (dump_overall_stats): Dump IPA porfile.
615 (dump_inline_stats): Dump IPA porfile.
616 (ipa_inline): Collect IPA stats.
617 * ipa-inline.h (clone_inlined_nodes): Update prototype.
618 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
619 (ipa_propagate_frequency): Use frequency function.
620 (ipa_profile): Cleanup.
621 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
622 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
623 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
624 (input_node): Do not stream frequency.
625 (input_edge): Do not stream frequency.
626 (merge_profile_summaries): Scale only IPA profiles.
627 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
628 * predict.c (drop_profile): Do not recompute frequency.
629 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
630 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
631 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
632 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
633 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
634 * tree-inline.c (copy_bb): Do not scale frequency.
635 (expand_call_inline): Do not scale frequency.
636 (tree_function_versioning): Do not scale frequency.
637 * ubsan.c (ubsan_create_edge): Do not pass frequency.
639 2017-11-10 Julia Koval <julia.koval@intel.com>
641 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8,
642 _mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8,
643 _mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8,
644 _mm256_maskz_gf2p8affine_epi64_epi8,
645 _mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8
646 _mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
647 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi,
648 __builtin_ia32_vgf2p8affineqb_v32qi,
649 __builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
650 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
652 2017-11-10 Tamar Christina <tamar.christina@arm.com>
656 (arm_option_override): Refactor.
657 (arm_option_reconfigure_globals): New.
658 (arm_options_perform_arch_sanity_checks): New.
659 * config/arm/arm-protos.h (arm_option_reconfigure_globals): New prototype.
660 (arm_options_perform_arch_sanity_checks): Likewise
662 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
664 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
666 * rs6000/rs6000.md ("type" attr): Add qmul.
667 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
668 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
669 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
671 2017-11-10 Martin Sebor <msebor@redhat.com>
674 * builtins.c (compute_objsize): Handle arrays that
675 compute_builtin_object_size likes to fail for. Make extern.
676 * builtins.h (compute_objsize): Declare.
677 (check_strncpy_sizes): New function.
678 (expand_builtin_strncpy): Call check_strncpy_sizes.
679 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
680 -Wstringop-truncation.
681 (gimple_fold_builtin_strncat): Same.
682 * gimple.c (gimple_build_call_from_tree): Set call location.
683 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
684 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
685 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
686 (handle_builtin_strlen): Use strlen_to_stridx.
687 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
689 Use strlen_to_stridx.
690 (pass_strlen::execute): Release strlen_to_stridx.
691 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
692 (-Wstringop-truncation): Document new option.
694 2017-11-10 Martin Liska <mliska@suse.cz>
696 PR gcov-profile/82702
697 * gcov.c (main): Handle intermediate files in a different
699 (get_gcov_intermediate_filename): New function.
700 (output_gcov_file): Remove support of intermediate files.
701 (generate_results): Allocate intermediate file.
702 (release_structures): Clean-up properly fn_end.
703 (output_intermediate_file): Start iterating with line 1.
705 2017-11-10 Jakub Jelinek <jakub@redhat.com>
707 PR tree-optimization/82929
708 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
709 ops_swapped_p non-static data member.
710 (store_immediate_info::store_immediate_info): Clear it.
711 (imm_store_chain_info::coalesce_immediate_stores): If swapping
712 ops set ops_swapped_p.
713 (count_multiple_uses): Handle ops_swapped_p.
715 2017-11-10 Martin Liska <mliska@suse.cz>
717 * coverage.c (coverage_init): Stream information about
718 support of has_unexecuted_blocks.
719 * doc/gcov.texi: Document that.
720 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
721 * gcov.c (read_graph_file): Likewise.
722 (output_line_beginning): Fix a small issue with
725 2017-11-10 Bin Cheng <bin.cheng@arm.com>
727 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
728 reference of trivial component.
730 2017-11-10 Jakub Jelinek <jakub@redhat.com>
733 * gimple-ssa-store-merging.c
734 (pass_store_merging::terminate_all_aliasing_chains): For
735 gimple_store_p stmts also call refs_output_dependent_p.
737 PR rtl-optimization/82913
738 * compare-elim.c (try_merge_compare): Punt if def_insn is not
741 2017-11-09 Jeff Law <law@redhat.com>
743 * vr-values.h: New file with vr_values class.
744 * tree-vrp.c: Include vr-values.h
745 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
746 data objects into the vr_values class.
747 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
748 (get_value_range): Make it a member function within vr_values class.
749 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
750 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
751 (op_with_constant_singleton_value_range): Likewise.
752 (extract_range_for_var_from_comparison_expr): Likewise.
753 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
754 (extract_range_from_binary_expr): Likewise.
755 (extract_range_from_unary_expr): Likewise.
756 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
757 (check_for_binary_op_overflow, extract_range_basic): Likewise.
758 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
759 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
760 (compare_name_with_value, compare_names): Likewise.
761 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
762 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
763 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
764 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
765 (extract_range_from_phi_node): Likewise.
766 (simplify_truth_ops_using_ranges): Likewise.
767 (simplify_div_or_mod_using_ranges): Likewise.
768 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
769 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
770 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
771 (simplify_float_conversion_using_ranges): Likewise.
772 (simplify_internal_call_using_ranges): Likewise.
773 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
774 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
775 poitner via x_vr_values for calls into gimple folder.
776 (vrp_initialize_lattice): Make this the vr_values ctor.
777 (vrp_free_lattice): Make this the vr_values dtor.
778 (set_vr_value): New function.
779 (class vrp_prop): Add vr_values data member. Add various member
780 functions as well as member functions that delegate to vr_values.
781 (check_array_ref): Make a member function within vrp_prop class.
782 (search_for_addr_array, vrp_initialize): Likewise.
783 (vrp_finalize): Likewise. Revamp to avoid direct access to
784 vr_value, values_propagated, etc.
785 (check_array_bounds): Extract vrp_prop class instance pointer from
786 walk info structure. Use it to call member functions.
787 (check_all_array_refs): Make a member function within vrp_prop class.
788 Smuggle class instance pointer via walk info structure.
789 (x_vr_values): New local static.
790 (vrp_valueize): Use x_vr_values to get class instance.
791 (vr_valueize_1): Likewise.
792 (class vrp_folder): Add vr_values data member. Add various member
793 functions as well as member functions that delegate to vr_values.
794 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
795 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
796 class instance from vr_values. Use it to call member functions.
797 (vrp_dom_walker): Add vr_values data member.
798 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
799 instance via x_vr_values.
800 (identify_jump_threads): Accept vr_values as argument. Store
801 it into the walker structure.
802 (evrp_dom_walker): Add vr_values class data member. Add various
804 (evrp_dom_walker::try_find_new_range): Use vr_values data
805 member to access the memory allocator.
806 (evrp_dom_walker::before_dom_children): Store vr_values class
807 instance into the vrp_folder class.
808 (evrp_dom_walker::push_value_range): Rework to avoid direct
809 access to num_vr_values and vr_value.
810 (evrp_dom_walker::pop_value_range): Likewise.
811 (execute_early_vrp): Remove call to vrp_initialize_lattice.
812 Use vr_values to get to dump_all_value_ranges member function.
813 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
814 and simplify_cond_using_ranges_2 via vrp_prop class instance.
815 Pass vr_values class instance down to identify_jump_threads.
816 Remove call to vrp_free_lattice.
817 (debug_all_value_ranges): Remove.
819 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
820 (vrp_folder): Likewise.
822 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
823 Get it from the existing bitmap instead.
824 (vrp_intersect_ranges_1): Likewise.
826 2017-11-09 Jakub Jelinek <jakub@redhat.com>
828 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
830 (store_immediate_info::store_immediate_info): Add bitnotp argument,
832 (imm_store_chain_info::coalesce_immediate_stores): Break group
833 if bit_not_p is different.
834 (count_multiple_uses, split_group,
835 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
836 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
837 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
838 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
839 to store_immediate_info ctor.
841 2017-11-09 Jim Wilson <jimw@sifive.com>
843 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
844 (scan_prog_file): Likewise.
846 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
848 * bb-reorder.c (max_entry_frequency): Remove.
849 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
850 connect_traces, push_to_next_round_p): Remove prototypes.
851 (find_traces_1_round): Use counts only.
852 (push_to_next_round_p): Likewise.
853 (find_traces): Likewise.
854 (rotate_loop): Likewise.
855 (find_traces_1_round): Likewise.
856 (connect_traces): Likewise.
857 (edge_order): Likewise.
859 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
861 * config/arm/arm.c (output_return_instruction): Add comments to
862 indicate requirement for cmse_nonsecure_entry return to account
863 for the size of clearing instruction output here.
864 (thumb_exit): Likewise.
865 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
866 return in hardfloat mode.
868 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
870 * config/rs6000/rs6000.c (machine_function): Add a bool,
871 "toc_is_wrapped_separately".
872 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
873 if it wasn't explicitly set or unset, we are optimizing for speed, and
874 doing separate shrink-wrapping.
875 (rs6000_get_separate_components): Enable the TOC component if
876 saving the TOC register in the prologue.
877 (rs6000_components_for_bb): Handle the TOC component.
878 (rs6000_emit_prologue_components): Store the TOC register where needed.
879 (rs6000_set_handled_components): Mark TOC as handled, if handled.
880 (rs6000_emit_prologue): Don't save the TOC if that is already done.
882 2017-11-09 Martin Jambor <mjambor@suse.cz>
884 * ipa-param-manipulation.c: New file.
885 * ipa-param-manipulation.h: Likewise.
886 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
887 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
888 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
889 (ipa_parm_adjustment): Likewise.
890 (ipa_parm_adjustment_vec): Likewise.
891 (ipa_get_vector_of_formal_parms): Moved declaration to
892 ipa-param-manipulation.h.
893 (ipa_get_vector_of_formal_parm_types): Likewise.
894 (ipa_modify_formal_parameters): Likewise.
895 (ipa_modify_call_arguments): Likewise.
896 (ipa_combine_adjustments): Likewise.
897 (ipa_dump_param_adjustments): Likewise.
898 (ipa_modify_expr): Likewise.
899 (ipa_get_adjustment_candidate): Likewise.
900 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
901 ipa-param-manipulation.c.
902 (ipa_get_vector_of_formal_parm_types): Likewise.
903 (ipa_modify_formal_parameters): Likewise.
904 (ipa_modify_call_arguments): Likewise.
905 (ipa_modify_expr): Likewise.
906 (get_ssa_base_param): Likewise.
907 (ipa_get_adjustment_candidate): Likewise.
908 (index_in_adjustments_multiple_times_p): Likewise.
909 (ipa_combine_adjustments): Likewise.
910 (ipa_dump_param_adjustments): Likewise.
911 * tree-sra.c: Also include ipa-param-manipulation.h
912 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
915 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
916 Alan Hayward <alan.hayward@arm.com>
917 David Sherwood <david.sherwood@arm.com>
919 * doc/sourcebuild.texi (vect_masked_store): Document.
921 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
922 Alan Hayward <alan.hayward@arm.com>
923 David Sherwood <david.sherwood@arm.com>
925 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
927 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
928 Alan Hayward <alan.hayward@arm.com>
929 David Sherwood <david.sherwood@arm.com>
931 * doc/sourcebuild.texi (vect_variable_length): Document.
933 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
934 Alan Hayward <alan.hayward@arm.com>
935 David Sherwood <david.sherwood@arm.com>
937 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
939 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
940 Alan Hayward <alan.hayward@arm.com>
941 David Sherwood <david.sherwood@arm.com>
943 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
945 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
946 Alan Hayward <alan.hayward@arm.com>
947 David Sherwood <david.sherwood@arm.com>
949 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
950 previously undocumented selectors.
951 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
953 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
955 * doc/rtl.texi (const_vector): Say that elements can be
957 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
958 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
959 (gen_vec_duplicate): Use it instead of CONSTANT_P.
960 * optabs.c (expand_vector_broadcast): Likewise.
962 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
963 Alan Hayward <alan.hayward@arm.com>
964 David Sherwood <david.sherwood@arm.com>
966 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
967 scaled index even if the unscaled address was invalid.
968 Don't increase the complexity of using a scale in that case.
970 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
971 Alan Hayward <alan.hayward@arm.com>
972 David Sherwood <david.sherwood@arm.com>
974 * doc/rtl.texi: Rewrite the subreg rules so that they partition
975 the inner register into REGMODE_NATURAL_SIZE bytes rather than
976 UNITS_PER_WORD bytes.
977 * emit-rtl.c (validate_subreg): Divide subregs into blocks
978 based on REGMODE_NATURAL_SIZE of the inner mode.
979 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
980 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
981 * expmed.c (lowpart_bit_field_p): Divide the value up into
982 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
983 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
984 whether something is likely to occupy more than one register.
986 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
989 * ipa-inline-transform.c (update_noncloned_frequencies): Use
990 profile_count::adjust_for_ipa_scaling.
991 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
992 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
994 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
996 2017-11-09 Jakub Jelinek <jakub@redhat.com>
998 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
999 (split_group): Add total_orig and total_new arguments, estimate the
1000 number of statements related to the store group without store merging
1001 and with store merging.
1002 (imm_store_chain_info::output_merged_store): Adjust split_group
1003 callers, punt if estimated number of statements with store merging
1004 is not smaller than estimated number of statements without it.
1006 (handled_load): Remove has_single_use checks.
1007 (pass_store_merging::process_store): Likewise.
1009 2017-11-09 Richard Biener <rguenther@suse.de>
1011 PR tree-optimization/82902
1012 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1014 2017-11-09 Martin Liska <mliska@suse.cz>
1017 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1020 2017-11-09 Martin Liska <mliska@suse.cz>
1022 PR tree-optimization/82669
1023 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1025 2017-11-09 Martin Liska <mliska@suse.cz>
1027 PR gcov-profile/48463
1028 * coverage.c (coverage_begin_function): Output also end locus
1029 of a function and information whether the function is
1031 * gcov-dump.c (tag_function): Parse and print the information.
1032 * gcov.c (INCLUDE_MAP): Add include.
1033 (INCLUDE_SET): Likewise.
1034 (struct line_info): Move earlier in the source file because
1035 of vector<line_info> in function_info structure.
1036 (line_info::line_info): Likewise.
1037 (line_info::has_block): Likewise.
1038 (struct source_info): Add new member index.
1039 (source_info::get_functions_at_location): New function.
1040 (function_info::group_line_p): New function.
1041 (output_intermediate_line): New function.
1042 (output_intermediate_file): Use the mentioned function.
1043 (struct function_start): New.
1044 (struct function_start_pair_hash): Likewise.
1045 (process_file): Add code that identifies group functions.
1046 Assign lines either to global or function scope.
1047 (generate_results): Skip artificial functions.
1048 (find_source): Assign index for each source file.
1049 (read_graph_file): Read new flag artificial and end_line.
1050 (add_line_counts): Assign it either to global of function scope.
1051 (accumulate_line_counts): Isolate core of the function to
1052 accumulate_line_info and call it for both function and global
1054 (accumulate_line_info): New function.
1055 (output_line_beginning): Fix GNU coding style.
1056 (print_source_line): New function.
1057 (output_line_details): Likewise.
1058 (output_function_details): Likewise.
1059 (output_lines): Iterate both source (global) scope and function
1061 (struct function_line_start_cmp): New class.
1062 * doc/gcov.texi: Reflect changes in documentation.
1064 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1067 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1068 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1069 and similarly for not instead of neg.
1071 2017-11-08 Andi Kleen <ak@linux.intel.com>
1073 * config/i386/i386.opt: Add -mforce-indirect-call.
1074 * config/i386/predicates.md: Check for flag_force_indirect_call.
1075 * doc/invoke.texi: Document -mforce-indirect-call
1077 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1079 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1081 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1082 config/riscv/riscv.c (predict.h): New include.
1083 (riscv_slow_unaligned_access_p): No longer static.
1084 (riscv_block_move_straight): Add require.
1085 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1087 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1090 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1091 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1092 nonimmediate_operand predicate for operand 1 instead of
1095 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1097 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1099 * config/aarch64/constraints.md (Uml): New constraint.
1100 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1103 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1105 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1106 of two vec_duplicates into a vec_concat.
1108 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1110 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1111 Simplify vec_merge of vec_duplicate and vec_concat.
1112 * config/aarch64/constraints.md (Utq): New constraint.
1113 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1116 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1118 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1119 Simplify vec_merge of vec_duplicate and const_vector.
1120 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1122 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1123 mode iterator. Update predicate on operand 1 to
1124 handle non-const_vec constants. Delete constraints.
1125 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1127 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1129 PR tree-optimization/78821
1130 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1132 (store_operand_info::store_operand_info): Initialize it to false.
1133 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1134 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1135 store in the group, and if chain_info is non-NULL, to ignore altogether
1137 (compatible_load_p): Fail if bit_not_p does not match.
1138 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1139 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1140 (pass_store_merging::process_store): Adjust
1141 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1142 call terminate_all_aliasing_chains newly when adding a store into
1143 a chain with non-NULL chain_info.
1145 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1147 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1149 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1151 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1153 (aarch64_layout_frame): Initialise emit_frame_chain.
1154 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1155 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1157 2017-11-08 Martin Liska <mliska@suse.cz>
1159 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1160 of gimple_call_internal_p.
1162 2017-11-07 Tom de Vries <tom@codesourcery.com>
1164 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1166 2017-11-07 Tom de Vries <tom@codesourcery.com>
1168 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1171 2017-11-08 Martin Liska <mliska@suse.cz>
1174 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1176 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1178 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1180 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1183 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1184 (*cmp<mode>_ccz_1): New insn with $k alternative.
1187 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1188 RTX_COMM_COMPARE as commutative as well.
1189 (ix86_binary_operator_ok): Formatting fix.
1190 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1191 *<code><mode>3<mask_name><round_saeonly_name>,
1192 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1193 *<s>mul<mode>3_highpart<mask_name>,
1194 *vec_widen_umult_even_v16si<mask_name>,
1195 *vec_widen_umult_even_v8si<mask_name>,
1196 *vec_widen_umult_even_v4si<mask_name>,
1197 *vec_widen_smult_even_v16si<mask_name>,
1198 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1199 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1200 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1201 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1202 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1203 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1204 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1205 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1206 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1207 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1208 ix86_binary_operator_ok. Formatting fixes.
1209 (*<plusminus_insn><mode>3<mask_name><round_name>,
1210 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1213 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1215 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1216 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1217 eq and ne if TARGET_ISEL.
1218 (cmp): New code_iterator.
1219 (UNS, UNSU_, UNSIK): New code_attrs.
1220 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1221 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1223 ("eq<mode>3"): ... this.
1224 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1226 ("ne<mode>3"): ... this.
1228 2017-11-07 Julia Koval <julia.koval@intel.com>
1231 * common/config/i386/i386-common.c
1232 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1233 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1234 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1235 * config/i386/i386.opt: Ditto.
1236 * config/i386/i386.c (ix86_target_string): Ditto.
1237 (ix86_option_override_internal): Ditto.
1238 (ix86_init_mpx_builtins): Move MPX to args2.
1239 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1240 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1241 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1242 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1243 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1244 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1245 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1247 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1250 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1251 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1252 (zero-extendsidi peephole2): Remove peephole.
1254 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1257 * stor-layout.c: Include attribs.h.
1258 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1259 explicit lookup of "aligned" attribute.
1261 2017-11-07 Andrew Waterman <andrew@sifive.com>
1263 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1264 (riscv_expand_block_move): Likewise.
1265 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1267 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1268 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1269 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1270 (riscv_adjust_block_mem): Likewise.
1271 (riscv_block_move_loop): Likewise.
1272 (riscv_expand_block_move): Likewise.
1273 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1275 2017-11-07 Michael Clark <michaeljclark@mac.com>
1277 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1278 (MUSL_DYNAMIC_LINKER): Likewise.
1280 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1282 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1283 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1286 2017-11-07 Richard Biener <rguenther@suse.de>
1288 * match.pd: Fix build.
1290 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1291 Jackson Woodruff <jackson.woodruff@arm.com>
1293 PR tree-optimization/71026
1294 * match.pd: Canonicalize negate in division.
1296 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1299 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1301 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1303 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1304 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1306 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1308 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1309 non-scalar integral types.
1310 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1311 (-(A-B), -(~A)): New transformations.
1313 2017-11-07 Tom de Vries <tom@codesourcery.com>
1315 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1316 semicolon after "do {} while (0)".
1317 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1318 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1319 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1320 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1321 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1323 2017-11-07 Tom de Vries <tom@codesourcery.com>
1325 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1326 after "do {} while (0)".
1327 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1328 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1329 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1330 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1331 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1333 2017-11-07 Tom de Vries <tom@codesourcery.com>
1336 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1338 (arm_rtx_costs_internal): Add missing semicolon after
1339 HANDLE_NARROW_SHIFT_ARITH call.
1341 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1343 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1344 disable isel if it was not set explicitly.
1346 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1348 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1350 (add_type_attribute) likewise.
1352 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1354 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1355 of struct ix86_frame.
1356 (ix86_initial_elimination_offset): Likewise.
1357 (ix86_expand_split_stack_prologue): Likewise.
1359 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1361 * tree-vrp.h (enum value_range_type): Update stale comment.
1363 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1364 Alan Hayward <alan.hayward@arm.com>
1365 David Sherwood <david.sherwood@arm.com>
1367 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1368 (aarch64_expand_vec_perm_const): Take the number of units too.
1369 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1370 (aarch64_expand_vec_perm_const): Likewise.
1371 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1372 (vec_perm<mode>): Update accordingly.
1374 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1375 Alan Hayward <alan.hayward@arm.com>
1376 David Sherwood <david.sherwood@arm.com>
1378 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1379 Take the number of units too.
1380 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1381 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1382 but check for a vector mode before rather than after the call.
1383 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1384 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1385 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1386 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1387 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1388 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1389 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1390 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1391 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1392 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1393 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1394 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1395 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1396 (aarch64_sqdmull2_n<mode>): Update accordingly.
1398 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1399 Alan Hayward <alan.hayward@arm.com>
1400 David Sherwood <david.sherwood@arm.com>
1402 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1403 the number of units too.
1404 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1405 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1406 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1407 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1408 (vec_store_lanesxi<mode>): Update accordingly.
1410 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1411 Alan Hayward <alan.hayward@arm.com>
1412 David Sherwood <david.sherwood@arm.com>
1414 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1415 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1416 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1417 of units rather than the mode.
1418 * config/aarch64/iterators.md (nunits): New mode attribute.
1419 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1420 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1421 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1422 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1423 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1424 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1425 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1426 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1427 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1428 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1429 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1430 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1431 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1432 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1433 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1434 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1435 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1436 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1437 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1438 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1439 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1440 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1441 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1442 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1443 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1444 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1445 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1446 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1447 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1448 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1449 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1450 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1451 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1452 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1453 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1454 (aarch64_simd_vec_setv2di): Likewise.
1456 2017-11-06 Carl Love <cel@us.ibm.com>
1458 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1460 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1461 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1463 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1465 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1466 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1467 Add power 8 macro expansions.
1468 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1469 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1470 power 8 instructions. (VSX_XXBR): Add iterator.
1472 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1474 * config/arm/arm.md (predicable_short_it): Change default to "no",
1475 improve documentation, remove uses that are identical to the default.
1476 (enabled_for_depr_it): Rename to enabled_for_short_it.
1477 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
1478 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1479 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1480 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1481 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1483 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1486 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1487 float128 helper macros, which are no longer used after deleting
1488 the old 'q' built-in functions, and moving the round to odd
1489 built-in functions to being special built-in functions.
1490 (BU_FLOAT128_2): Likewise.
1491 (BU_FLOAT128_1_HW): Likewise.
1492 (BU_FLOAT128_2_HW): Likewise.
1493 (BU_FLOAT128_3_HW): Likewise.
1494 (FABSQ): Delete old 'q' built-in functions.
1495 (COPYSIGNQ): Likewise.
1496 (SQRTF128_ODD): Move round to odd built-in functions to be
1497 special built-in functions, so that we can handle
1498 -mabi=ieeelongdouble.
1499 (TRUNCF128_ODD): Likewise.
1500 (ADDF128_ODD): Likewise.
1501 (SUBF128_ODD): Likewise.
1502 (MULF128_ODD): Likewise.
1503 (DIVF128_ODD): Likewise.
1504 (FMAF128_ODD): Likewise.
1505 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1506 built-in names to 'f128'.
1507 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1508 old 'q' built-in functions, as the machine independent code for
1509 'f128' built-in functions handles this.
1510 (rs6000_expand_builtin): Add expansion for float128 round to odd
1511 functions, keying off on -mabi=ieeelongdouble of whether to use
1512 the KFmode or TFmode variant.
1513 (rs6000_init_builtins): Initialize the _Float128 round to odd
1515 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1516 _Float128 'q' built-in functions are now mapped into the new
1517 'f128' built-in functions.
1519 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1521 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1522 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1524 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1527 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1528 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1529 * toplev.c: Include "ipa-fnsummary.h".
1530 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1532 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1534 PR tree-optimization/82838
1535 * gimple-ssa-store-merging.c
1536 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1537 on a separate gimple_seq which is then appended to seq.
1539 2017-11-06 Jeff Law <law@redhat.com>
1542 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1543 (get_probe_interval): New functions.
1544 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1545 (ix86_adjust_stack_and_probe): Likewise.
1546 (output_adjust_stack_and_probe): Likewise.
1547 (ix86_emit_probe_stack_range): Likewise.
1548 (ix86_expand_prologue): Likewise.
1550 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1552 PR tree-optimization/82816
1553 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1554 if the modes of the two types are the same.
1555 (convert_plusminus_to_widen): Likewise.
1557 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1559 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
1561 (usadv16qi): New define_expand.
1562 (usadv8hi): New define_expand.
1564 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
1567 * ipa-inline-transform.c (update_noncloned_frequencies): Always
1569 (inline_transform): Likewise.
1570 * predict.c (counts_to_freqs): Remove useless conditional.
1571 * profile-count.h (profile_count::apply_scale): Move sanity check.
1572 * tree-inline.c (copy_bb): Always scale.
1573 (copy_cfg_body): Likewise.
1575 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
1578 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
1580 (*cmp_ite0): Add enabled_for_depr_it attribute.
1581 (*cmp_ite1): Likewise.
1583 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1585 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
1588 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1590 * tree-vrp.c (vrp_int_const_binop): Return true on success and
1591 return the value by pointer.
1592 (extract_range_from_multiplicative_op_1): Update accordingly.
1593 Return as soon as an operation fails.
1595 2017-11-05 Tom de Vries <tom@codesourcery.com>
1598 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
1599 (DEF_SANITIZER_BUILTIN): ... here.
1600 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
1601 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
1603 2017-11-05 Tom de Vries <tom@codesourcery.com>
1606 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
1608 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
1609 ASM_OUTPUT_BEFORE_CASE_LABEL call.
1610 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
1612 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1613 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1614 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1616 2017-11-05 Tom de Vries <tom@codesourcery.com>
1619 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
1622 2017-11-04 Michael Clark <michaeljclark@mac.com>
1624 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
1625 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
1627 (*addsi3_extended): Likewise.
1628 (*addsi3_extended2): Likewise.
1629 (<optab>si3): Likewise.
1630 (<optab>di3): Likewise.
1631 (<optab><mode>3): Likewise.
1632 (<*optabe>si3_internal): Likewise.
1633 (zero_extendqi<SUPERQI:mode>2): Likewise.
1634 (*add<mode>hi3): Likewise.
1635 (*xor<mode>hi3): Likewise.
1636 (<optab>di3): Likewise.
1637 (*<optab>si3_extend): Likewise.
1638 (*sge<u>_<X:mode><GPR:mode>): Likewise.
1639 (*slt<u>_<X:mode><GPR:mode>): Likewise.
1640 (*sle<u>_<X:mode><GPR:mode>): Likewise.
1642 2017-11-04 Andrew Waterman <andrew@sifive.com>
1644 * config/riscv/riscv.c (riscv_option_override): Conditionally set
1645 TARGET_STRICT_ALIGN based upon -mtune argument.
1647 2017-11-04 Andrew Waterman <andrew@sifive.com>
1649 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
1651 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
1653 config/i386/i386.c (choose_basereg): Use optional scratch
1654 register and add assertion.
1655 (x86_emit_outlined_ms2sysv_save): Use scratch register when
1656 needed, and don't allocate stack.
1657 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
1658 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
1659 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
1661 2017-11-03 Jeff Law <law@redhat.com>
1663 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
1664 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
1665 to probe at the start of a noreturn function.
1667 2017-11-03 Jakub Jelinek <jakub@redhat.com>
1669 PR tree-optimization/78821
1670 * gimple-ssa-store-merging.c: Update the file comment.
1671 (MAX_STORE_ALIAS_CHECKS): Define.
1672 (struct store_operand_info): New type.
1673 (store_operand_info::store_operand_info): New constructor.
1674 (struct store_immediate_info): Add rhs_code and ops data members.
1675 (store_immediate_info::store_immediate_info): Add rhscode, op0r
1676 and op1r arguments to the ctor, initialize corresponding data members.
1677 (struct merged_store_group): Add load_align_base and load_align
1679 (merged_store_group::merged_store_group): Initialize them.
1680 (merged_store_group::do_merge): Update them.
1681 (merged_store_group::apply_stores): Pick the constant for
1682 encode_tree_to_bitpos from one of the two operands, or skip
1683 encode_tree_to_bitpos if neither operand is a constant.
1684 (class pass_store_merging): Add process_store method decl. Remove
1685 bool argument from terminate_all_aliasing_chains method decl.
1686 (pass_store_merging::terminate_all_aliasing_chains): Remove
1687 var_offset_p argument and corresponding handling.
1688 (stmts_may_clobber_ref_p): New function.
1689 (compatible_load_p): New function.
1690 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
1691 if there is overlap and rhs_code is not INTEGER_CST. For
1692 non-overlapping stores terminate group if rhs is not mergeable.
1693 (get_alias_type_for_stmts): Change first argument from
1694 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
1695 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
1696 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
1698 (get_location_for_stmts): Change first argument from
1699 auto_vec<gimple *> & to vec<gimple *> &.
1700 (struct split_store): Remove orig_stmts data member, add orig_stores.
1701 (split_store::split_store): Create orig_stores rather than orig_stmts.
1702 (find_constituent_stmts): Renamed to ...
1703 (find_constituent_stores): ... this. Change second argument from
1704 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
1705 to info structures rather than the statements.
1706 (split_group): Rename ALLOW_UNALIGNED argument to
1707 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
1708 it. Adjust find_constituent_stores caller.
1709 (imm_store_chain_info::output_merged_store): Handle rhs_code other
1710 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
1711 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
1712 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
1713 (mem_valid_for_store_merging): New function.
1714 (handled_load): New function.
1715 (pass_store_merging::process_store): New method.
1716 (pass_store_merging::execute): Use process_store method. Adjust
1717 terminate_all_aliasing_chains caller.
1719 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1721 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
1722 Return true for more constants, symbols and label references.
1723 (aarch64_valid_floating_const): Remove unused function.
1725 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1728 * config/aarch64/aarch64.c (aarch64_layout_frame):
1729 Undo forcing of LR at bottom of frame.
1731 2017-11-03 Jeff Law <law@redhat.com>
1734 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
1735 for int_registers_saved.
1737 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
1738 extracted from tree-ssa-dom.c.
1739 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1740 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1741 (record_equivalences_from_incoming_edge): Add additional argument
1742 to single_pred_edge_ignoring_loop_edges call.
1743 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1744 (uncprop_dom_walker::before_dom_children): Add additional argument
1745 to single_pred_edge_ignoring_loop_edges call.
1746 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1747 single_pred_edge_ignoring_loop_edges rather than open coding.
1748 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1750 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1752 * match.pd (-(-A)): Rewrite.
1754 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1756 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1757 (rs6000_emit_int_cmove): New declaration.
1758 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1759 (rs6000_emit_sISEL): Delete.
1760 (rs6000_emit_int_cmove): Make non-static.
1761 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1762 instead of rs6000_emit_sISEL.
1764 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1766 * asan.c (create_cond_insert_point): Maintain profile.
1767 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
1769 * basic-block.h (struct basic_block_def): Remove frequency.
1770 (EDGE_FREQUENCY): Use to_frequency
1771 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
1773 (find_traces): Update to use to_frequency.
1774 (find_traces_1_round): Likewise; use only IPA counts.
1775 (bb_to_key): Likewise.
1776 (connect_traces): Use IPA counts only.
1777 (copy_bb_p): Update to use to_frequency.
1778 (fix_up_crossing_landing_pad): Likewise.
1779 (sanitize_hot_paths): Likewise.
1780 * bt-load.c (basic_block_freq): Likewise.
1781 * cfg.c (init_flow): Set count_max to uninitialized.
1782 (check_bb_profile): Remove frequencies; check counts.
1783 (dump_bb_info): Do not dump frequencies.
1784 (update_bb_profile_for_threading): Update counts only.
1785 (scale_bbs_frequencies_int): Likewise.
1786 (MAX_SAFE_MULTIPLIER): Remove.
1787 (scale_bbs_frequencies_gcov_type): Update counts only.
1788 (scale_bbs_frequencies_profile_count): Update counts only.
1789 (scale_bbs_frequencies): Update counts only.
1790 * cfg.h (struct control_flow_graph): Add count-max.
1791 (update_bb_profile_for_threading): Update prototype.
1792 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
1793 (find_many_sub_basic_blocks): Likewise.
1794 * cfgcleanup.c (try_forward_edges): Likewise.
1795 (try_crossjump_to_edge): Likewise.
1796 * cfgexpand.c (expand_gimple_cond): Likewise.
1797 (expand_gimple_tailcall): Likewise.
1798 (construct_init_block): Likewise.
1799 (construct_exit_block): Likewise.
1800 * cfghooks.c (verify_flow_info): Check consistency of counts.
1801 (dump_bb_for_graph): Do not dump frequencies.
1802 (split_block_1): Do not update frequencies.
1803 (split_edge): Do not update frequencies.
1804 (make_forwarder_block): Do not update frequencies.
1805 (duplicate_block): Do not update frequencies.
1806 (account_profile_record): Do not update frequencies.
1807 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
1808 for global heuristics.
1809 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
1810 (expected_loop_iterations_unbounded): Use counts only.
1811 * cfgloopmanip.c (scale_loop_profile): Simplify.
1812 (create_empty_loop_on_edge): Simplify
1814 (duplicate_loop_to_header_edge): Simplify
1815 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
1816 (update_br_prob_note): Take care of removing note when profile
1818 (relink_block_chain): Do not dump frequency.
1819 (rtl_account_profile_record): Use to_frequency.
1820 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1821 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1822 (cgraph_update_edges_for_call_stmt_node): Likewise.
1823 (cgraph_edge::verify_count_and_frequency): Update.
1824 (cgraph_node::verify_node): Temporarily disable frequency verification.
1825 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1826 to_cgraph_frequency.
1827 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1828 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1830 (cgraph_node::expand_thunk): Update profile.
1831 * except.c (dw2_build_landing_pads): Do not update frequency.
1832 * final.c (compute_alignments): Use to_frequency.
1833 (dump_basic_block_info): Do not dump frequency.
1834 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1835 (dump_gimple_bb_header): Do not dump frequency.
1836 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1838 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1839 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1840 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1841 (init_before_recovery): Do not update frequency.
1842 (sched_create_recovery_edges): Do not update frequency.
1843 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1844 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1845 (ipa_cp_c_finalize): Set max_count to uninitialized.
1846 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1847 (param_change_prob): Use counts.
1848 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1850 * ipa-split.c (consider_split): Use to_frequency.
1851 (split_function): Use to_frequency.
1852 * ira-build.c (loop_compare_func): Likewise.
1853 (mark_loops_for_removal): Likewise.
1854 (mark_all_loops_for_removal): Likewise.
1855 * loop-doloop.c (doloop_modify): Do not update frequency.
1856 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1858 * lto-streamer-in.c (input_function): Update count_max.
1859 * omp-expand.c (expand_omp_taskreg): Update count_max.
1860 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1861 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1862 (maybe_hot_count_p): Use ipa counts only.
1863 (maybe_hot_bb_p): Simplify.
1864 (maybe_hot_edge_p): Simplify.
1865 (probably_never_executed): Do not take frequency argument.
1866 (probably_never_executed_bb_p): Do not pass frequency.
1867 (probably_never_executed_edge_p): Likewise.
1868 (combine_predictions_for_bb): Check that profile is nonzero.
1869 (propagate_freq): Do not set frequency.
1870 (drop_profile): Simplify.
1871 (counts_to_freqs): Simplify.
1872 (expensive_function_p): Use to_frequency.
1873 (propagate_unlikely_bbs_forward): Simplify.
1874 (determine_unlikely_bbs): Simplify.
1875 (estimate_bb_frequencies): Add hack to silence graphite issues.
1876 (compute_function_frequency): Use ipa counts.
1877 (pass_profile::execute): Update.
1878 (rebuild_frequencies): Use counts only.
1879 (force_edge_cold): Use counts only.
1880 * profile-count.c (profile_count::dump): Dump new count types.
1881 (profile_count::differs_from_p): Check compatiblity.
1882 (profile_count::to_frequency): New function.
1883 (profile_count::to_cgraph_frequency): New function.
1884 * profile-count.h (struct function): Declare.
1885 (enum profile_quality): Add profile_guessed_local and
1886 profile_guessed_global0.
1887 (class profile_proability): Decrease number of bits to 29;
1888 update from_reg_br_prob_note and to_reg_br_prob_note.
1889 (class profile_count: Update comment; decrease number of bits
1890 to 61. Check compatibility.
1891 (profile_count::compatible_p): New private member function.
1892 (profile_count::ipa_p): New member function.
1893 (profile_count::operator<): Handle global zero correctly.
1894 (profile_count::operator>): Handle global zero correctly.
1895 (profile_count::operator<=): Handle global zero correctly.
1896 (profile_count::operator>=): Handle global zero correctly.
1897 (profile_count::nonzero_p): New member function.
1898 (profile_count::force_nonzero): New member function.
1899 (profile_count::max): New member function.
1900 (profile_count::apply_scale): Handle IPA scalling.
1901 (profile_count::guessed_local): New member function.
1902 (profile_count::global0): New member function.
1903 (profile_count::ipa): New member function.
1904 (profile_count::to_frequency): Declare.
1905 (profile_count::to_cgraph_frequency): Declare.
1906 * profile.c (OVERLAP_BASE): Delete.
1907 (compute_frequency_overlap): Delete.
1908 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1909 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1910 * sched-ebb.c (rank): Use counts only.
1911 * shrink-wrap.c (handle_simple_exit): Use counts only.
1912 (try_shrink_wrapping): Use counts only.
1913 (place_prologue_for_one_component): Use counts only.
1914 * tracer.c (find_best_predecessor): Use to_frequency.
1915 (find_trace): Use to_frequency.
1916 (tail_duplicate): Use to_frequency.
1917 * trans-mem.c (expand_transaction): Do not update frequency.
1918 * tree-call-cdce.c: Do not update frequency.
1919 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1920 (gimple_merge_blocks): Likewise.
1921 (gimple_split_edge): Likewise.
1922 (gimple_duplicate_sese_region): Likewise.
1923 (gimple_duplicate_sese_tail): Likewise.
1924 (move_sese_region_to_fn): Likewise.
1925 (gimple_account_profile_record): Likewise.
1926 (insert_cond_bb): Likewise.
1927 * tree-complex.c (expand_complex_div_wide): Likewise.
1928 * tree-eh.c (lower_resx): Update profile.
1929 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1931 (initialize_cfun): Do not initialize frequencies
1932 (freqs_to_counts): Delete.
1933 (copy_cfg_body): Ignore count parameter.
1934 (copy_body): Update.
1935 (expand_call_inline): Update count_max.
1936 (optimize_inline_calls): Update count_max.
1937 (tree_function_versioning): Update count_max.
1938 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1939 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1941 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1942 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1943 (try_peel_loop): Likewise.
1944 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1946 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1947 (tree_transform_and_unroll_loop): Do not use frequencies
1948 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1949 Use reliable prediction only.
1950 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1951 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1952 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1953 probability scaling.
1954 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1955 not update frequency
1956 (any_remaining_duplicated_blocks): Likewise.
1957 (update_profile): Likewise.
1958 (estimated_freqs_path): Delete.
1959 (freqs_to_counts_path): Delete.
1960 (clear_counts_path): Delete.
1961 (ssa_fix_duplicate_block_edges): Likewise.
1962 (duplicate_thread_path): Likewise.
1963 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1964 * tree-tailcall.c (decrease_profile): Do not update frequency.
1965 (eliminate_tail_call): Likewise.
1966 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1967 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1968 (optimize_mask_stores): Likewise.
1969 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1970 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1971 (ubsan_expand_ptr_ifn): Update profile.
1972 * value-prof.c (gimple_ic): Simplify.
1973 * value-prof.h (gimple_ic): Update prototype.
1974 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1975 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1977 (want_inline_self_recursive_call_p): Likewise.
1978 (resolve_noninline_speculation): Only cummulate defined counts.
1979 (inline_small_functions): Use nonzero_p.
1980 (ipa_inline): Do not access freed node.
1982 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1984 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1985 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1987 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1989 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1990 non-legitimate address.
1992 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1994 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1995 (*lt0_<mode>di, *lt0_<mode>si): New.
1997 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1999 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2000 TARGET_PAIRED_FLOAT.
2002 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2003 Jim Wilson <jim.wilson@linaro.org>
2005 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2006 * config/aarch64/aarch64-tune.md: Regenerated.
2007 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2008 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2010 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2012 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2013 (arc_expand_prologue): Restore blink for millicode.
2014 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2016 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2019 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2020 gen_vec_duplicate after forcing the scalar into a register.
2022 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2024 * combine (try_combine): Print the insns input to try_combine to the
2027 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2030 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2031 Remove second argument from aarch64_process_target_attr call.
2032 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2034 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2036 (aarch64_handle_attr_arch): Remove second argument.
2037 (aarch64_handle_attr_cpu): Ditto.
2038 (aarch64_handle_attr_tune): Ditto.
2039 (aarch64_handle_attr_isa_flags): Ditto.
2040 (aarch64_process_one_target_attr): Ditto.
2041 (aarch64_process_target_attr): Ditto.
2042 (aarch64_option_valid_attribute_p): Remove second argument.
2043 on aarch64_process_target_attr call.
2045 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2047 * diagnostic.c: Include "selftest-diagnostic.h".
2048 (selftest::assert_location_text): New function.
2049 (selftest::test_diagnostic_get_location_text): New function.
2050 (selftest::diagnostic_c_tests): Call it.
2052 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2054 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2055 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2056 (class selftest::test_diagnostic_context): Move to...
2057 * selftest-diagnostic.c: New file.
2058 * selftest-diagnostic.h: New file.
2060 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2062 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2063 offset range for FT32B.
2064 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2065 * config/ft32/ft32.md: Add TARGET_NOPM.
2066 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2067 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2069 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2071 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2073 2017-11-02 Jeff Law <law@redhat.com>
2075 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2076 virtual keyword on FINAL OVERRIDE members.
2078 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2079 virtuals together. Add virtual destructor.
2080 (substitute_and_fold_engine): Similarly.
2082 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2084 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2086 2017-11-02 Richard Biener <rguenther@suse.de>
2088 PR tree-optimization/82795
2089 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2091 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2093 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2095 (gcc_GAS_CHECK_FEATURE): Remove.
2096 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2097 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2098 * configure: Regenerate.
2100 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2102 * config/arc/arc.c (hwloop_optimize): Account for empty
2105 2017-11-02 Richard Biener <rguenther@suse.de>
2108 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2109 Truncate ARRAY_REF index and element size.
2111 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2113 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2115 2017-11-01 Jeff Law <law@redhat.com>
2117 * tree-ssa-ccp.c (ccp_folder): New class derived from
2118 substitute_and_fold_engine.
2119 (ccp_folder::get_value): New member function.
2120 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2121 (ccp_fold_stmt): Remove prototype.
2122 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2123 * tree-ssa-copy.c (copy_folder): New class derived from
2124 substitute_and_fold_engine.
2125 (copy_folder::get_value): Renamed from get_value.
2126 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2127 * tree-vrp.c (vrp_folder): New class derived from
2128 substitute_and_fold_engine.
2129 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2130 (vrp_folder::get_value): New member function.
2131 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2132 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2133 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2134 provide a class interface to folder/substitute routines.
2135 (ssa_prop_fold_stmt_fn): Remove typedef.
2136 (ssa_prop_get_value_fn): Likewise.
2137 (subsitute_and_fold): Remove prototype.
2138 (replace_uses_in): Likewise.
2139 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2140 Renamed from replace_uses_in. Call the virtual member function
2141 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2142 (substitute_and_fold_dom_walker): Remove initialization of
2143 data member entries for calbacks. Add substitute_and_fold_engine
2144 member and initialize it.
2145 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2146 member functions for get_value, replace_phi_args_in c
2147 replace_uses_in, and fold_stmt calls.
2148 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2149 substitute_and_fold. Remove assert. Update ctor call.
2151 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2152 (ssa_prop_visit_phi_fn): Likewise.
2153 (class ssa_propagation_engine): New class to provide an interface
2155 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2157 (ssa_prop_visit_phi): Likewise.
2158 (ssa_propagation_engine::simulate_stmt): Moved into class.
2159 Call visit_phi/visit_stmt from the class rather than via
2160 file scoped static variables.
2161 (ssa_propagation_engine::simulate_block): Moved into class.
2162 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2163 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2164 set file scoped statics for the visit_stmt/visit_phi callbacks.
2165 * tree-complex.c (complex_propagate): New class derived from
2166 ssa_propagation_engine.
2167 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2168 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2169 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2171 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2172 ssa_propagation_engine.
2173 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2174 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2175 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2176 * tree-ssa-copy.c (copy_prop): New class derived from
2177 ssa_propagation_engine.
2178 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2179 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2180 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2181 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2182 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2183 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2184 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2186 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2188 PR rtl-optimization/82778
2189 PR rtl-optimization/82597
2190 * compare-elim.c (struct comparison): Add in_a_setter field.
2191 (find_comparison_dom_walker::before_dom_children): Remove killed
2192 bitmap and df_simulate_find_defs call, instead walk the defs.
2193 Compute last_setter and initialize in_a_setter. Merge definitions
2194 with first initialization for a few variables.
2195 (try_validate_parallel): Use insn_invalid_p instead of
2196 recog_memoized. Return insn rather than just the pattern.
2197 (try_merge_compare): Fix up comment. Don't uselessly test if
2198 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2200 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2203 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2204 Alan Hayward <alan.hayward@arm.com>
2205 David Sherwood <david.sherwood@arm.com>
2207 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2208 aarch64_hard_regno_nregs to get the number of registers
2211 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2212 Alan Hayward <alan.hayward@arm.com>
2213 David Sherwood <david.sherwood@arm.com>
2215 * config/aarch64/constraints.md (Upl): Rename to...
2217 * config/aarch64/aarch64.md
2218 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2221 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2222 Alan Hayward <alan.hayward@arm.com>
2223 David Sherwood <david.sherwood@arm.com>
2225 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2226 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2229 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2230 Alan Hayward <alan.hayward@arm.com>
2231 David Sherwood <david.sherwood@arm.com>
2233 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2234 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2235 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2237 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2238 of a single element.
2239 * config/aarch64/iterators.md: Add a comment above the permute
2240 unspecs to say that they are generated directly by
2241 aarch64_expand_vec_perm_const.
2242 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2244 2017-11-01 Nathan Sidwell <nathan@acm.org>
2246 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2248 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2250 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2251 and medany code models, and describe what they do.
2253 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2255 Revert accidental duplicate:
2257 * combine.c (can_change_dest_mode): Reject changes in
2258 REGMODE_NATURAL_SIZE.
2260 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2262 PR rtl-optimization/64682
2263 PR rtl-optimization/69567
2264 PR rtl-optimization/69737
2265 PR rtl-optimization/82683
2266 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2267 register mentioned in the note, drop the note, unless it came from I3,
2268 in which case it should go to I3 again.
2270 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2272 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2273 and return false if not.
2274 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2276 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2278 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2279 range is known to be empty.
2281 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2282 Alan Hayward <alan.hayward@arm.com>
2283 David Sherwood <david.sherwood@arm.com>
2285 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2286 and CONST_VECTOR_NUNITS instead of computing the number of units from
2287 the byte sizes of the vector and element.
2288 (simplify_binary_operation_1): Likewise.
2289 (simplify_const_binary_operation): Likewise.
2290 (simplify_ternary_operation): Likewise.
2292 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2293 Alan Hayward <alan.hayward@arm.com>
2294 David Sherwood <david.sherwood@arm.com>
2296 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2297 (int_mem_offset): ...this new function.
2298 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2299 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2300 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2303 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2304 Alan Hayward <alan.hayward@arm.com>
2305 David Sherwood <david.sherwood@arm.com>
2307 * lower-subreg.c (interesting_mode_p): New function.
2308 (compute_costs, find_decomposable_subregs, decompose_register)
2309 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2310 (resolve_clobber, dump_choices): Use it.
2312 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2313 Alan Hayward <alan.hayward@arm.com>
2314 David Sherwood <david.sherwood@arm.com>
2316 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2318 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2319 Alan Hayward <alan.hayward@arm.com>
2320 David Sherwood <david.sherwood@arm.com>
2322 * alias.c (find_base_value, find_base_term): Only process integer
2323 truncations. Check the precision rather than the size.
2325 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2326 Alan Hayward <alan.hayward@arm.com>
2327 David Sherwood <david.sherwood@arm.com>
2329 * machmode.h (is_narrower_int_mode): New function
2330 * optabs.c (expand_float, expand_fix): Use it.
2331 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2333 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2334 Alan Hayward <alan.hayward@arm.com>
2335 David Sherwood <david.sherwood@arm.com>
2337 * rtl.h (narrower_subreg_mode): New function.
2338 * ira-color.c (update_costs_from_allocno): Use it.
2340 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2341 Alan Hayward <alan.hayward@arm.com>
2342 David Sherwood <david.sherwood@arm.com>
2344 * optabs-query.h (convert_optab_p): New function, split out from...
2345 (convert_optab_handler): ...here.
2346 (widening_optab_handler): Delete.
2347 (find_widening_optab_handler): Remove permit_non_widening parameter.
2348 (find_widening_optab_handler_and_mode): Likewise. Provide an
2349 override that operates on mode class wrappers.
2350 * optabs-query.c (widening_optab_handler): Delete.
2351 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2352 parameter. Assert that the two modes are the same class and that
2353 the "from" mode is narrower than the "to" mode. Use
2354 convert_optab_handler instead of widening_optab_handler.
2355 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2356 instead of widening_optab_handler.
2357 * expr.c (expand_expr_real_2): Update calls to
2358 find_widening_optab_handler.
2359 * optabs.c (expand_widen_pattern_expr): Likewise.
2360 (expand_binop_directly): Take the insn_code as a parameter.
2361 (expand_binop): Only call find_widening_optab_handler for
2362 conversion optabs; use optab_handler otherwise. Update calls
2363 to find_widening_optab_handler and expand_binop_directly.
2364 Use convert_optab_handler instead of widening_optab_handler.
2365 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2366 find_widening_optab_handler and use scalar_mode rather than
2368 (convert_plusminus_to_widen): Likewise.
2370 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2371 Alan Hayward <alan.hayward@arm.com>
2372 David Sherwood <david.sherwood@arm.com>
2374 * machmode.h (fixed_size_mode): New class.
2375 * rtl.h (get_pool_mode): Return fixed_size_mode.
2376 * gengtype.c (main): Add fixed_size_mode.
2377 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2378 (get_raw_arg_mode): Likewise.
2379 * doc/tm.texi: Regenerate.
2380 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2381 * targhooks.c (default_get_reg_raw_mode): Likewise.
2382 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2383 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2384 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2385 (msp430_get_raw_result_mode): Likewise.
2386 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2387 * dbxout.c (dbxout_parms): Require fixed-size modes.
2388 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2389 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2390 * omp-low.c (lower_oacc_reductions): Likewise.
2391 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2392 (simplify_subreg): Update accordingly.
2393 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2394 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2395 that aren't fixed-size.
2396 (get_pool_mode): Return a fixed_size_mode.
2397 (output_constant_pool_2): Take a fixed_size_mode.
2399 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2400 Alan Hayward <alan.hayward@arm.com>
2401 David Sherwood <david.sherwood@arm.com>
2403 * doc/rtl.texi (vec_series): Document.
2404 (const): Say that the operand can be a vec_series.
2405 * rtl.def (VEC_SERIES): New rtx code.
2406 * rtl.h (const_vec_series_p_1): Declare.
2407 (const_vec_series_p): New function.
2408 * emit-rtl.h (gen_const_vec_series): Declare.
2409 (gen_vec_series): Likewise.
2410 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2411 (gen_vec_series): Likewise.
2412 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2413 * simplify-rtx.c (simplify_unary_operation): Handle negations
2415 (simplify_binary_operation_series): New function.
2416 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2417 (test_vector_ops_series): New function.
2418 (test_vector_ops): Call it.
2419 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2420 gen_const_vec_series.
2421 (altivec_lvsr): Likewise.
2422 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2424 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2425 Alan Hayward <alan.hayward@arm.com>
2426 David Sherwood <david.sherwood@arm.com>
2428 * doc/rtl.texi (const): Update description of address constants.
2429 Say that vector constants are allowed too.
2430 * common.md (E, F): Use CONSTANT_P instead of checking for
2432 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2433 checking for CONST_VECTOR.
2434 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2436 * expr.c (expand_expr_real_2): Check for vector modes instead
2437 of checking for CONST_VECTOR.
2438 * rtl.h (const_vec_p): New function.
2439 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2440 (unwrap_const_vec_duplicate): Handle them here too.
2442 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2443 David Malcolm <dmalcolm@redhat.com>
2444 Alan Hayward <alan.hayward@arm.com>
2445 David Sherwood <david.sherwood@arm.com>
2447 * rtl.h (vec_duplicate_p): New function.
2448 * selftest-rtl.c (assert_rtx_eq_at): New function.
2449 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2450 (assert_rtx_eq_at): Declare.
2451 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2452 * selftest-run-tests.c (selftest::run_tests): Call it.
2453 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2454 (simplify_unary_operation_1): Recursively handle vector duplicates.
2455 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2457 (simplify_subreg): Handle subregs of vector duplicates.
2458 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2459 (selftest::simplify_rtx_c_tests): New functions.
2461 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2462 Alan Hayward <alan.hayward@arm.com>
2463 David Sherwood <david.sherwood@arm.com>
2465 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2466 (gen_vec_duplicate): Likewise.
2467 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2469 (gen_const_vector): ...here.
2470 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2471 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2472 whose elements are all equal.
2473 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2474 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2475 (simplify_relational_operation): Likewise.
2476 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2478 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2479 (aarch64_expand_vector_init): Likewise.
2480 * config/arm/arm.c (neon_vdup_constant): Likewise.
2481 (neon_expand_vector_init): Likewise.
2482 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2483 (arm_block_set_unaligned_vect): Likewise.
2484 (arm_block_set_aligned_vect): Likewise.
2485 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2486 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2487 (expand_vec_perm_even_odd_pack): Likewise.
2488 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2489 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2490 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2491 gen_const_vec_duplicate.
2492 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2493 * config/mips/mips.c (mips_gen_const_int_vector): Use
2494 gen_const_vec_duplicate.
2495 (mips_expand_vector_init): Use CONST0_RTX.
2496 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2497 (define_split): Use gen_const_vec_duplicate.
2498 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2499 (define_split): Use gen_const_vec_duplicate.
2500 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2501 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2502 * config/spu/spu.c (spu_const): Likewise.
2504 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2505 Alan Hayward <alan.hayward@arm.com>
2506 David Sherwood <david.sherwood@arm.com>
2508 * combine.c (can_change_dest_mode): Reject changes in
2509 REGMODE_NATURAL_SIZE.
2511 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2513 * configure.ac (--enable-libssp): New.
2514 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2515 trying to determine target-specific default. Adjust indentation.
2516 * configure: Regenerated.
2517 * doc/install.texi (Configuration): Expand --disable-libssp
2520 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2522 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2525 2017-10-31 Martin Jambor <mjambor@suse.cz>
2528 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2530 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2532 * auto-profile.c (autofdo_source_profile::read): Use
2533 UNKNOWN_LOCATION rather than 0.
2534 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2535 (warning_at): ...this overload.
2536 (warning_at_rich_loc_n): Rename to...
2537 (warning_n): ...this overload.
2538 (error_at_rich_loc): Rename to...
2539 (error_at): ...this overload.
2540 (pedwarn_at_rich_loc): Rename to...
2541 (pedwarn): ...this overload.
2542 (permerror_at_rich_loc): Rename to...
2543 (permerror): ...this overload.
2544 (inform_at_rich_loc): Rename to...
2545 (inform): ...this overload.
2546 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2547 (diagnostic_n_impl_richloc): Rename to...
2548 (diagnostic_n_impl): ...this rich_location *-based decl.
2549 (inform_at_rich_loc): Rename to...
2550 (inform): ...this, and add an assertion.
2551 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2552 (warning_at_rich_loc): Rename to...
2553 (warning_at): ...this, and add an assertion.
2554 (warning_at_rich_loc_n): Rename to...
2555 (warning_n): ...this, and add an assertion.
2556 (warning_n): Update location_t-based implementation for removal of
2557 location_t-based diagnostic_n_impl.
2558 (pedwarn_at_rich_loc): Rename to...
2559 (pedwarn): ...this, and add an assertion.
2560 (permerror_at_rich_loc): Rename to...
2561 (permerror): ...this, and add an assertion.
2562 (error_n): Update for removal of location_t-based diagnostic_n_impl.
2563 (error_at_rich_loc): Rename to...
2564 (error_at): ...this, and add an assertion.
2565 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
2566 (driver::do_spec_on_infiles): Likewise.
2567 * substring-locations.c (format_warning_va): Update for renaming
2568 of inform_at_rich_loc.
2570 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2572 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
2573 _Float<N>X built-in functions so that the variant without the
2574 "__builtin_" prefix is only enabled for the GNU C and Objective C
2575 languages when they are in non-strict ANSI/ISO mode.
2576 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
2577 * target.def (floatn_builtin_p): Add a target hook to control
2578 whether _Float<N> and _Float<N>X built-in functions without the
2579 "__builtin_" prefix are enabled, and return true for C and
2580 Objective C in the default hook. Include langhooks.h in
2582 * targhooks.h (default_floatn_builtin_p): Likewise.
2583 * targhooks.c (default_floatn_builtin_p): Likewise.
2584 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
2585 floatn_builtin_p target hook.
2586 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
2588 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
2589 Eric Botcazou <ebotcazou@adacore.com>
2591 PR rtl-optimization/81803
2592 * lra-constraints.c (curr_insn_transform): Also reload the whole
2593 register for a strict subreg no wider than a word if this is for
2594 a WORD_REGISTER_OPERATIONS target.
2596 2017-10-31 Jason Merrill <jason@redhat.com>
2598 * gdbinit.in: Skip over inlines from timevar.h.
2600 2017-10-31 Martin Liska <mliska@suse.cz>
2602 * doc/gcov.texi: Document new option.
2603 * gcov.c (print_usage): Likewise print it.
2604 (process_args): Support the argument.
2605 (format_count): New function.
2606 (format_gcov): Use the function.
2608 2017-10-31 Martin Liska <mliska@suse.cz>
2610 * gcov.c (struct name_map): do not use typedef.
2611 Define operator== and operator<.
2612 (name_search): Remove.
2613 (name_sort): Remove.
2614 (main): Do not allocate names.
2615 (process_file): Add vertical space.
2616 (generate_results): Use std::find.
2617 (release_structures): Do not release memory.
2618 (find_source): Use std::find.
2620 2017-10-31 Martin Liska <mliska@suse.cz>
2622 * gcov.c (struct line_info): Remove it's typedef.
2623 (line_info::line_info): Add proper ctor.
2624 (line_info::has_block): Do not use a typedef.
2625 (struct source_info): Do not use typedef.
2626 (circuit): Likewise.
2627 (get_cycles_count): Likewise.
2628 (output_intermediate_file): Iterate via vector iterator.
2629 (add_line_counts): Use std::vector methods.
2630 (accumulate_line_counts): Likewise.
2631 (output_lines): Likewise.
2633 2017-10-31 Martin Liska <mliska@suse.cz>
2635 * gcov.c (struct source_info): Remove typedef.
2636 (source_info::source_info): Add proper ctor.
2637 (accumulate_line_counts): Use struct, not it's typedef.
2638 (output_gcov_file): Likewise.
2639 (output_lines): Likewise.
2640 (main): Do not allocate an array.
2641 (output_intermediate_file): Use size of vector container.
2642 (process_file): Resize the vector.
2643 (generate_results): Do not preallocate, use newly added vector
2645 (release_structures): Do not release sources.
2646 (find_source): Use vector methods.
2647 (add_line_counts): Do not use typedef.
2649 2017-10-31 Martin Liska <mliska@suse.cz>
2651 * doc/gcov.texi: Document that.
2652 * gcov.c (add_line_counts): Mark lines with a non-executed
2654 (output_line_beginning): Handle such lines.
2655 (output_lines): Pass new argument.
2656 (output_intermediate_file): Print it in intermediate format.
2658 2017-10-31 Martin Liska <mliska@suse.cz>
2660 * color-macros.h: New file.
2661 * diagnostic-color.c: Factor out color related to macros to
2663 * doc/gcov.texi: Document -k option.
2664 * gcov.c (INCLUDE_STRING): Include string.h.
2665 (print_usage): Add -k option.
2666 (process_args): Parse it.
2667 (pad_count_string): New function.
2668 (output_line_beginning): Likewise.
2669 (DEFAULT_LINE_START): New macro.
2670 (output_lines): Support color output.
2672 2017-10-31 Martin Liska <mliska@suse.cz>
2674 PR gcov-profile/82633
2675 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
2676 their interaction with GCOV infrastructure.
2677 * configure.ac: Add -fkeep-{inline,static}-functions to
2679 * configure: Regenerate.
2681 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
2684 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
2686 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
2689 * config/rs6000/rs6000.md (allocate_stack): Force update interval
2690 into a register if it does not fit into an immediate offset field.
2692 2017-10-31 Olivier Hainque <hainque@adacore.com>
2694 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
2696 2017-10-31 Julia Koval <julia.koval@intel.com>
2698 * config.gcc: Add gfniintrin.h.
2699 * config/i386/gfniintrin.h: New.
2700 * config/i386/i386-builtin-types.def
2701 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2702 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2703 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2704 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2705 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2706 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
2707 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
2708 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
2709 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
2710 V64QI_FTYPE_V64QI_V64QI_INT): New types.
2711 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2712 * config/i386/immintrin.h: Include gfniintrin.h.
2713 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
2715 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
2717 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
2719 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2721 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
2722 (arm_ashldi3_1bit): Remove pattern.
2723 (ashrdi3): Remove shift by 1 expansion.
2724 (arm_ashrdi3_1bit): Remove pattern.
2725 (lshrdi3): Remove shift by 1 expansion.
2726 (arm_lshrdi3_1bit): Remove pattern.
2727 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
2728 cost of ashldi3 by 1.
2729 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
2730 (<shift>di3_neon): Likewise.
2732 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2734 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
2735 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
2736 and (*aarch64_simd_mov<VQ:mode>).
2737 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
2738 pattern alternative.
2739 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2740 attributes to match pattern alternative.
2742 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2744 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2745 * config/rs6000/emmintrin.h: New file.
2746 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2748 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2750 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2751 * (movdi_vfp_cortexa8): Remove pattern.
2753 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2755 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2757 (Specific, alpha*-dec-osf5.1): Remove.
2758 (Specific, mips-sgi-irix5): Remove.
2759 (Specific, mips-sgi-irix6): Remove.
2761 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2764 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2765 arguments to clear_bit_region_be.
2767 2017-10-30 Jim Wilson <wilson@tuliptree.org>
2769 * gimplify.c: Include memmodel.h.
2771 2017-10-30 Martin Jambor <mjambor@suse.cz>
2773 * omp-grid.c (grid_attempt_target_gridification): Also insert a
2774 condition whether loop should be executed at all.
2776 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
2778 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2779 gimple folding of vec_madd() intrinsics.
2780 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
2781 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
2782 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
2784 2017-10-30 Richard Biener <rguenther@suse.de>
2786 PR tree-optimization/82762
2788 2017-10-23 Richard Biener <rguenther@suse.de>
2790 PR tree-optimization/82129
2792 2017-08-01 Richard Biener <rguenther@suse.de>
2794 PR tree-optimization/81181
2795 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2796 (compute_antic): ... end of iteration here.
2798 2017-10-30 Joseph Myers <joseph@codesourcery.com>
2800 * doc/invoke.texi (C Dialect Options): Document -std=c17,
2801 -std=iso9899:2017 and -std=gnu17.
2802 * doc/standards.texi (C Language): Document C17 support.
2803 * doc/cpp.texi (Overview): Mention -std=c17.
2804 (Standard Predefined Macros): Document C11 and C17 values of
2805 __STDC_VERSION__. Do not refer to C99 support as incomplete.
2806 * doc/extend.texi (Inline): Do not list individual options for
2807 standards newer than C99.
2808 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
2810 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
2813 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2815 * asan.c (asan_finish_file): Align asan globals array by shadow
2818 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2821 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2822 (struct store_immediate_info): Add bitregion_start and bitregion_end
2824 (store_immediate_info::store_immediate_info): Add brs and bre
2825 arguments and initialize bitregion_{start,end} from those.
2826 (struct merged_store_group): Add bitregion_start, bitregion_end,
2827 align_base and mask fields. Drop unnecessary struct keyword from
2828 struct store_immediate_info. Add do_merge method.
2829 (clear_bit_region_be): Use memset instead of loop storing zeros.
2830 (merged_store_group::do_merge): New method.
2831 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2832 stores as long as the surrounding bitregions have no gaps.
2833 (merged_store_group::merge_overlapping): Use do_merge.
2834 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2835 is byte aligned, rather than requiring that start and width are
2836 byte aligned. Drop unnecessary struct keyword from
2837 struct store_immediate_info. Allocate and populate also mask array.
2838 Make start of the arrays relative to bitregion_start rather than
2839 start and size them according to bitregion_{end,start} difference.
2840 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2841 struct store_immediate_info.
2842 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2843 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2844 struct keyword from struct store_immediate_info.
2845 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2846 between stores as long as the surrounding bitregions have no gaps.
2848 (struct split_store): Add orig non-static data member.
2849 (split_store::split_store): Initialize orig to false.
2850 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2851 if there is exactly a single original stmt. Change stmts argument
2852 to pointer from reference, if NULL, don't push anything to it. Add
2853 first argument, use it to optimize skipping over orig stmts that
2854 are known to be before bitpos already. Simplify.
2855 (split_group): Return unsigned int count how many stores are or
2856 would be needed rather than a bool. Add allow_unaligned argument.
2857 Change split_stores argument from reference to pointer, if NULL,
2858 only do a dry run computing how many stores would be produced.
2859 Rewritten algorithm to use both alignment and misalign if
2860 !allow_unaligned and handle bitfield stores with gaps.
2861 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2862 from bitregion_start instead of start. Compute allow_unaligned
2863 here, if true, do 2 split_group dry runs to compute which one
2864 produces fewer stores and prefer aligned if equal. Punt if
2865 new count is bigger or equal than original before emitting any
2866 statements, rather than during that. Remove no longer needed
2867 new_ssa_names tracking. Replace num_stmts with
2868 split_stores.length (). Use 32-bit stack allocated entries
2869 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2870 if possible. Handle bitfields with gaps.
2871 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2872 Compute bitregion_{start,end} for the stores and construct
2873 store_immediate_info with that. Formatting fixes.
2875 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2878 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2879 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2881 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2883 * gimplify.c: Include tm_p.h.
2885 * common.opt (gcoff): Re-add as ignored option.
2886 (gcoff1, gcoff2, gcoff3): Likewise.
2888 * Makefile.in (OBJS): Delete sdbout.o.
2889 (GTFILES): Delete $(srcdir)/sdbout.c.
2890 * debug.h: Delete sdb_debug_hooks.
2891 * final.c: Delete sdbout.h include.
2892 (final_scan_insn): Delete SDB_DEBUG check.
2893 (rest_of_clean_state): Likewise.
2894 * output.h: Delete sdb_begin_function_line.
2897 * toplev.c: Delete sdbout.h include.
2898 (process_options): Delete SDB_DEBUG check.
2899 * tree-core.h (tree_type_common): Delete pointer field of
2901 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2902 TYPE_SYMTAB_POINTER.
2903 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2904 (TYPE_SYMTAB_IS_POINTER): Delete.
2905 (TYPE_SYMTAB_IS_DIE): Renumber.
2906 * xcoffout.c: Refer to former sdbout.c file.
2907 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2909 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2910 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2911 (Debugging Options): Delete -gcoff.
2912 (-gstabs): Delete SDB reference.
2914 (-gcoff@var{level}): Delete.
2915 * doc/passes.texi (Debugging information output): Delete SDB and
2916 sdbout.c references.
2917 * doc/tm.texi: Regenerate.
2918 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2919 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2921 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2922 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2923 (SDB_DEBUGGING_INFO): Delete.
2924 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2925 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2926 * target.def (output_source_filename): Delete COFF reference.
2928 * common.opt (gcoff): Delete.
2929 (gxcoff+): Update Negative chain.
2930 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2932 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2933 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2934 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2936 (expand_function_start): Change sdb reference to past tense.
2937 (expand_function_end): Change sdb reference to past tense.
2938 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2939 * opts.c (debug_type_names): Delete coff entry.
2940 (common_handle_option): Delete OPT_gcoff case.
2941 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2943 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2944 * config/cris/cris.h: Delete SDB reference in comment.
2945 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2946 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2947 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2948 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2950 (ix86_expand_prologue): Likewise.
2951 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2952 * config/ia64/ia64.h: Likewise.
2953 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2954 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2956 * config/mmix/mmix.h: Likewise.
2957 * config/nds32/nds32.c: Likewise.
2958 * config/stormy/storym16.h: Likewise.
2959 * config/visium/visium.h: Likewise.
2960 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2962 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2964 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2965 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2966 FRAME_POINTER_REGNUM point at high end of local var area.
2968 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2970 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2971 Move comment around. Do not reset best_edge for a copiable
2972 destination if the copy would cause a partition change.
2973 (better_edge_p): Remove redundant check.
2975 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2977 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2979 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2981 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2982 for math functions that have _Float<N> and _Float<N>X variants.
2983 (mathfn_built_in_2): Add support for math functions that have
2984 _Float<N> and _Float<N>X variants.
2985 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2986 (expand_builtin_mathfn_ternary): Add support for fma with
2987 _Float<N> and _Float<N>X variants.
2988 (expand_builtin): Likewise.
2989 (fold_builtin_3): Likewise.
2990 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2991 create math function _Float<N> and _Float<N>X variants as external
2993 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2994 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2995 the __builtin_ prefix and if not strict ansi, without the prefix.
2996 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2997 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2998 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2999 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3000 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3001 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3002 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3003 function signatures for fma _Float<N> and _Float<N>X variants.
3004 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3005 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3006 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3007 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3008 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3009 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3010 * gencfn-macros.c (print_case_cfn): Add support for math functions
3011 that have _Float<N> and _Float<N>X variants.
3012 (print_define_operator_list): Likewise.
3013 (fltfn_suffixes): Likewise.
3015 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3016 for math functions that have _Float<N> and _Float<N>X variants.
3017 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3018 and _Float<N>X variants.
3019 (COPYSIGN): Likewise.
3022 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3023 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3025 (integer_valued_read_call_p): Likewise.
3026 * fold-const-call.c (fold_const_call_ss): Likewise.
3027 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3028 _Float<N> and _Float<N>X variants.
3029 (fold_const_call_ssss): Add support for fma _Float<N> and
3030 _Float<N>X variants.
3031 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3032 support for copysign and fma _Float<N> and _Float<N>X variants.
3033 (backprop::process_builtin_call_use): Likewise.
3034 * tree-call-cdce.c (can_test_argument_range); Add support for
3035 sqrt _Float<N> and _Float<N>X variants.
3036 (edom_only_function): Likewise.
3037 (get_no_error_domain): Likewise.
3038 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3039 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3040 copysign _Float<N> and _Float<N>X variants.
3041 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3042 handled by machine independent code.
3043 (FMAF128): Likewise.
3044 * doc/cpp.texi (Common Predefined Macros): Document defining
3045 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3046 fma _Float<N> and _Float<N>X variants.
3048 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3051 * config/i386/i386-modes.def (CCFPU): Remove definition.
3052 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3053 (ix86_cc_modes_compatible): Ditto.
3054 (ix86_expand_carry_flag_compare): Ditto.
3055 (ix86_expand_int_movcc): Ditto.
3056 (ix86_expand_int_addcc): Ditto.
3057 (ix86_reverse_condition): Ditto.
3058 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3059 Return true/false for unordered/ordered fp comparisons.
3060 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3061 (ix86_prepare_fp_compare_args): Update for rename.
3062 (ix86_expand_fp_compare): Update for rename. Generate unordered
3063 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3064 (ix86_expand_sse_compare_and_jump): Ditto.
3065 * config/i386/predicates.md (fcmov_comparison_operator):
3066 Remove CCFPU mode handling.
3067 (ix86_comparison_operator): Ditto.
3068 (ix86_carry_flag_operator): Ditto.
3069 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3070 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3071 (*cmpu<mode>_cc_i387): Ditto.
3072 (FPCMP): Remove mode iterator.
3073 (unord): Remove mode attribute.
3074 (unord_subst): New define_subst transformation
3075 (unord): New define_subst attribute.
3077 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3078 (*cmpi<unord>xf_i387): Ditto.
3079 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3080 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3081 using unord_subst transformation.
3082 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3083 (round_saeonly): Also handle CCFP mode.
3084 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3085 Remove UNSPEC_SAHF unspec handling.
3087 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3089 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3091 2017-10-27 Jeff Law <law@redhat.com>
3093 * gimple-ssa-sprintf.c: Include domwalk.h.
3094 (class sprintf_dom_walker): New class, derived from dom_walker.
3095 (sprintf_dom_walker::before_dom_children): New function.
3096 (struct call_info): Moved into sprintf_dom_walker class
3097 (compute_formath_length, handle_gimple_call): Likewise.
3098 (sprintf_length::execute): Call the dominator walker rather
3099 than walking the statements.
3101 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3102 gimple statement locations.
3103 (check_array_bounds): Corresponding changes. Get the statement's
3104 location directly from wi->stmt.
3106 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3109 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3111 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3113 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3114 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3116 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3119 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3120 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3121 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3122 maybe_get_pool_constant.
3123 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3126 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3128 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3129 2.26 caveat. Update gas and gld versions.
3130 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3133 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3135 * cgraph.h (set_malloc_flag): Declare.
3136 * cgraph.c (set_malloc_flag_1): New function.
3137 (set_malloc_flag): Likewise.
3138 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3139 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3141 (read_ipa_call_summary): Add support for reading is_return_callee.
3142 (write_ipa_call_summary): Stream is_return_callee.
3143 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3144 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3145 ipa-prop.h, ipa-fnsummary.h.
3146 (pure_const_names): Change to static.
3147 (malloc_state_e): Define.
3148 (malloc_state_names): Define.
3149 (funct_state_d): Add field malloc_state.
3150 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3151 (check_retval_uses): New function.
3152 (malloc_candidate_p): Likewise.
3153 (analyze_function): Add support for malloc attribute.
3154 (pure_const_write_summary): Stream malloc_state.
3155 (pure_const_read_summary): Add support for reading malloc_state.
3156 (dump_malloc_lattice): New function.
3157 (propagate_malloc): New function.
3158 (warn_function_malloc): New function.
3159 (ipa_pure_const::execute): Call propagate_malloc and
3160 ipa_free_fn_summary.
3161 (pass_local_pure_const::execute): Add support for malloc attribute.
3162 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3163 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3165 2017-10-27 Martin Liska <mliska@suse.cz>
3167 PR gcov-profile/82457
3168 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3169 for fork-like functions to be properly instrumented.
3171 2017-10-27 Richard Biener <rguenther@suse.de>
3174 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3175 info when we redirected EH.
3177 2017-10-26 Michael Collison <michael.collison@arm.com>
3179 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3181 (<optab>_trunchf<GPI:mode>2: New pattern.
3182 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3183 * config/aarch64/iterators.md (wv): New mode attribute.
3184 (vf, VF): New mode attributes.
3185 (vgp, VGP): New mode attributes.
3186 (s): Update attribute with SImode and DImode prefixes.
3188 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3190 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3191 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3192 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3193 (nios2_option_overide): Initialize it. Don't allow R0-relative
3194 addressing with PIC.
3195 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3196 (nios2_symbolic_constant_p): Likewise.
3197 (nios2_legitimate_address_p): Likewise.
3198 (nios2_r0rel_section_name_p): New.
3199 (nios2_symbol_ref_in_r0rel_data_p): New.
3200 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3201 (r0rel_constant_p): New.
3202 (nios2_print_operand_address): Handle r0rel_constant_p.
3203 (nios2_cdx_narrow_form_p): Likewise.
3204 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3205 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3206 (Nios II Options): Document -mr0rel-sec.
3208 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3210 * config/nios2/nios2.c: Include xregex.h.
3211 (nios2_gprel_sec_regex): New.
3212 (nios2_option_overide): Initialize it. Don't allow GP-relative
3213 addressing with PIC.
3214 (nios2_small_section_name_p): Check for regex match.
3215 * config/nios2/nios2.opt (mgprel-sec=): New option.
3216 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3217 (Nios II Options): Document -mgprel-sec.
3219 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3221 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3223 2017-10-26 Tom de Vries <tom@codesourcery.com>
3225 PR tree-optimization/82707
3226 * gimple.c (gimple_copy): Fix unsharing of
3227 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3229 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3231 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3232 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3233 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3234 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3235 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3236 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3237 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3238 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3239 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3240 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3241 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3242 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3243 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3244 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3245 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3246 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3247 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3249 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3251 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3253 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3254 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3255 warning to rs6000.c. Remove the Undocumented flag, since it has
3257 (-mabi=ibmlongdouble): Likewise.
3258 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3259 already set, set the default format for long double.
3260 (rs6000_debug_reg_global): Print whether long double is IBM or
3262 (rs6000_option_override_internal): Rework setting long double
3263 format. Only warn if the user is changing the long double default
3264 and they did not use -Wno-psabi.
3265 * doc/invoke.texi (PowerPC options): Update the documentation for
3266 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3268 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3269 Alan Hayward <alan.hayward@arm.com>
3270 David Sherwood <david.sherwood@arm.com>
3272 * rtl.h (wider_subreg_mode): New function.
3273 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3274 rather than an unsigned int *.
3275 * ira-color.c (regno_max_ref_width): Replace with...
3276 (regno_max_ref_mode): ...this new variable.
3277 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3278 Use wider_subreg_mode.
3279 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3280 rather than an unsigned int *.
3281 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3282 (process_alt_operands): Likewise.
3283 (invariant_p): Likewise.
3284 * lra-spills.c (assign_mem_slot): Likewise.
3285 (add_pseudo_to_slot): Likewise.
3286 * lra.c (collect_non_operand_hard_regs): Likewise.
3287 (add_regs_to_insn_regno_info): Likewise.
3288 * reload1.c (regno_max_ref_width): Replace with...
3289 (regno_max_ref_mode): ...this new variable.
3290 (reload): Update accordingly. Update call to
3291 ira_sort_regnos_for_alter_reg.
3292 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3293 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3294 (scan_paradoxical_subregs): Likewise.
3296 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3298 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3299 (aarch64_frame): Add emit_frame_chain boolean.
3300 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3301 Move eh_return case to aarch64_layout_frame.
3302 (aarch64_layout_frame): Initialize emit_frame_chain.
3303 (aarch64_expand_prologue): Use emit_frame_chain.
3305 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3307 * config/aarch64/aarch64.c (aarch64_layout_frame):
3308 Ensure LR is always stored at the bottom of the callee-saves.
3309 Remove rarely used frame layout which saves callee-saves at top of
3310 frame, so the store of LR can be used as a valid probe in all cases.
3312 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3314 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3315 Improve unaligned TImode/TFmode base/offset split.
3317 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3318 Alan Hayward <alan.hayward@arm.com>
3319 David Sherwood <david.sherwood@arm.com>
3321 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3322 * combine.c (find_single_use_1): Likewise.
3323 (expand_field_assignment): Likewise.
3324 (move_deaths): Likewise.
3325 * lra-constraints.c (simplify_operand_subreg): Likewise.
3326 (curr_insn_transform): Likewise.
3327 * lra.c (collect_non_operand_hard_regs): Likewise.
3328 (add_regs_to_insn_regno_info): Likewise.
3329 * rtlanal.c (reg_referenced_p): Likewise.
3330 (covers_regno_no_parallel_p): Likewise.
3332 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3334 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3335 Don't print any bits outside the precision of the value.
3336 * wide-int.cc (test_printing): Add some new tests.
3338 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3340 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3341 supports -xbrace_comment option.
3342 * configure: Regenerate.
3343 * config.in: Regenerate.
3344 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3345 (ASM_CPU_SPEC): Use it.
3347 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3349 * target.def (static_rtx_alignment): New hook.
3350 * targhooks.h (default_static_rtx_alignment): Declare.
3351 * targhooks.c (default_static_rtx_alignment): New function.
3352 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3353 * doc/tm.texi: Regenerate.
3354 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3355 instead of targetm.constant_alignment. Remove call to
3357 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3358 (cris_preferred_mininum_alignment): New function, split out from...
3359 (cris_constant_alignment): ...here.
3360 (cris_static_rtx_alignment): New function.
3361 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3363 (ix86_constant_alignment): ...here.
3364 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3365 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3366 (mmix_static_rtx_alignment): New function.
3367 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3368 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3370 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3373 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3374 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3376 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3378 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3379 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3380 mark as a sign-extending load.
3381 (local_pic_load_u): Define.
3383 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3386 * fold-const.c (operand_equal_for_comparison_p): Also return true
3387 if ARG0 is a simple variant of ARG1 with narrower precision.
3388 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3390 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3392 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3394 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3395 scatter_static, scatter_per_elt.
3396 * x86-tune-costs.h: Add new cost entries.
3398 2017-10-25 Richard Biener <rguenther@suse.de>
3400 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3401 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3402 class pass_fre): Move to ...
3403 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3404 class pass_fre): ... here and adjust for statistics.
3406 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3409 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3410 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3411 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3412 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3415 2017-10-25 Richard Biener <rguenther@suse.de>
3417 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3418 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3419 eliminate_push_avail, eliminate_insert): Move inside...
3420 (class eliminate_dom_walker): ... this class in preparation
3422 (fini_eliminate): Remove by merging with ...
3423 (eliminate): ... this function. Adjust for class changes.
3424 (pass_pre::execute): Remove fini_eliminate call.
3425 (pass_fre::execute): Likewise.
3427 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3430 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3431 (VPERMI2, VPERMI2I): New mode iterators.
3432 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3433 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3435 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3436 mode iterator. Remove 3 old define_insn patterns.
3437 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3438 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3439 VPERMI2 mode iterator, remove the other two expanders.
3440 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3441 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3442 instructions, remove the other two patterns.
3443 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3444 mode iterator, remove the other two patterns.
3445 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3446 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3447 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3448 and adjust argument order accordingly.
3449 (ix86_expand_vec_perm): Adjust caller.
3450 (expand_vec_perm_1): Likewise.
3451 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3452 (expand_vec_perm_vpermt2_vpshub2): ... this.
3453 (ix86_expand_vec_perm_const_1): Adjust caller.
3454 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3457 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3458 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3459 (vec_shl_<mode>): Remove unused expander.
3460 (avx512bw_<shift_insn><mode>3): New define_insn.
3461 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3462 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3464 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3467 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3470 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3472 PR rtl-optimization/82396
3473 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3474 (autopref_multipass_init): Simplify initialization.
3475 (autopref_rank_data): Simplify sort order.
3476 * gcc/sched-int.h (autopref_multipass_data_): Remove
3477 multi_mem_insn_p, min_offset and max_offset.
3479 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3482 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3483 Check special value of flag_omit_frame_pointer.
3484 (aarch64_can_eliminate): Likewise.
3485 (aarch64_override_options_after_change_1): Simplify handling of
3486 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3488 2017-10-24 Richard Biener <rguenther@suse.de>
3490 PR tree-optimization/82697
3491 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3492 zero for conditional load and unconditional store.
3494 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3496 * doc/install.texi: Document bootstrap-cet.
3498 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3501 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3502 ENDBR instruction at function entrance if function is only
3505 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3508 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3509 patterns to better describe from which operation the CF is computed.
3510 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3511 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3512 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3513 is 0, use _0 suffixed expanders instead of emitting a comparison
3516 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3518 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3519 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3521 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3524 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3525 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3526 * loop-iv.c (iv_get_reaching_def): Likewise.
3527 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3528 variable is promoted and the partition contains undefined values.
3530 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3532 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3534 (nios2_address_cost): Define.
3535 (nios2_legitimize_address): Recognize (exp + constant) directly.
3536 (TARGET_ADDRESS_COST): Define.
3538 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3540 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3541 (nios2_symbolic_memory_operand_p): Declare.
3542 (nios2_split_large_constant): Declare.
3543 (nios2_split_symbolic_memory_operand): Declare.
3544 * config/nios2/nios2.c: Adjust includes.
3545 (nios2_symbolic_constant_allowed): New.
3546 (nios2_symbolic_constant_p): New.
3547 (nios2_plus_symbolic_constant_p): New.
3548 (nios2_valid_addr_expr_p): Recognize addresses involving
3550 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3551 (nios2_symbolic_memory_operand_p): New.
3552 (nios2_large_constant_p): New.
3553 (nios2_split_large_constant): New.
3554 (nios2_split_plus_large_constant): New.
3555 (nios2_split_symbolic_memory_operand): New.
3556 (nios2_legitimize_address): Code refactoring. Handle addresses
3557 involving symbolic constants.
3558 (nios2_emit_move_sequence): Likewise.
3559 (nios2_print_operand): Improve error output.
3560 (nios2_print_operand_address): Handle LO_SUM.
3561 (nios2_cdx_narrow_form_p): Likewise.
3562 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
3563 operands involving symbolic constants.
3564 (movhi_internal, movsi_internal): Likewise.
3565 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
3566 (extendhisi2, extendqi<mode>2): Likewise.
3568 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3570 * tree-pass.h (PROP_rtl_split_insns): Define.
3571 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
3573 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3575 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
3577 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3580 * target.def (const_not_ok_for_debug_p): Default to
3581 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
3582 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
3583 * targhooks.c (default_const_not_ok_for_debug_p): New function.
3584 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
3585 which targetm.const_not_ok_for_debug_p returned true.
3586 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
3588 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
3590 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
3591 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
3592 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
3594 (ix86_const_not_ok_for_debug_p): New function.
3595 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
3596 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
3598 2017-10-23 David Malcolm <dmalcolm@redhat.com>
3601 * system.h: Conditionally include "unique-ptr.h" if
3602 INCLUDE_UNIQUE_PTR is defined.
3603 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
3604 of defining INCLUDE_UNIQUE_PTR before including "system.h".
3606 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
3608 * config/rl78/rl78.md: New define_expand "subdi3".
3610 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
3613 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
3614 DF_REF_INSN if DF_REF_INSN_INFO is false.
3616 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
3618 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
3619 xmm_move instead of sse_move.
3620 (sse_store_index): New function.
3621 (ix86_register_move_cost): Be more sensible about mismatch stall;
3622 model AVX moves correctly; make difference between sse->integer and
3624 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
3625 moves; make difference between SSE and AVX.
3626 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
3627 and zmm_move. Increase size of sse load and store tables;
3628 add unaligned load and store tables; add ssemmx_to_integer.
3629 * x86-tune-costs.h: Update all entries according to real
3630 move latencies from Agner Fog's manual and chip documentation.
3632 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3635 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
3636 * config/i386/constraints.md (Wf): New constraint.
3637 * config/i386/i386.md (UNSPEC_SBB): New unspec.
3638 (cmp<dwi>_doubleword): Removed.
3639 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
3640 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
3641 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
3642 expand with cmp<dwi>_doubleword. For LTU and GEU use
3643 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
3645 * common.opt (gcolumn-info): Enable by default.
3646 * doc/invoke.texi (gcolumn-info): Document new default.
3648 2017-10-23 Richard Biener <rguenther@suse.de>
3650 PR tree-optimization/82672
3651 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
3652 Fold the stmt if we propagated into it.
3654 2017-10-23 Richard Biener <rguenther@suse.de>
3656 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
3657 (bitmap_remove_expr_from_set): ... this. All callers call this
3658 for non-constant values.
3659 (bitmap_set_subtract): Rename to...
3660 (bitmap_set_subtract_expressions): ... this. Adjust and
3662 (bitmap_set_contains_value): Remove superfluous check.
3663 (bitmap_set_replace_value): Inline into single caller ...
3664 (bitmap_value_replace_in_set): ... here and simplify.
3665 (dependent_clean): Merge into ...
3666 (clean): ... this using an overload. Adjust.
3667 (prune_clobbered_mems): Adjust.
3668 (compute_antic_aux): Likewise.
3669 (compute_partial_antic_aux): Likewise.
3671 2017-10-23 Richard Biener <rguenther@suse.de>
3673 PR tree-optimization/82129
3675 2017-08-01 Richard Biener <rguenther@suse.de>
3677 PR tree-optimization/81181
3678 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3679 (compute_antic): ... end of iteration here.
3681 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3683 * target.def (starting_frame_offset): New hook.
3684 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
3685 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
3686 * doc/tm.texi.in: Regenerate.
3687 * hooks.h (hook_hwi_void_0): Declare.
3688 * hooks.c (hook_hwi_void_0): New function.
3689 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
3690 STARTING_FRAME_OFFSET.
3691 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
3692 * reload1.c (reload): Likewise.
3693 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
3694 instead of STARTING_FRAME_OFFSET.
3695 * function.c (try_fit_stack_local): Likewise.
3696 (assign_stack_local_1): Likewise
3697 (instantiate_virtual_regs): Likewise.
3698 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
3699 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
3700 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
3701 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
3702 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
3703 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
3704 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
3705 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
3706 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
3707 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
3708 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
3709 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
3710 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
3711 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
3712 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
3713 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
3714 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
3715 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
3716 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
3717 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
3718 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
3719 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
3720 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
3721 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
3722 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
3723 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
3724 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
3725 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
3726 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
3727 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
3728 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
3729 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
3730 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
3731 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
3732 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
3733 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
3734 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
3735 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
3736 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
3737 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
3738 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
3739 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3740 return a HOST_WIDE_INT.
3741 (avr_builtin_setjmp_frame_value): Use it instead of
3742 STARTING_FRAME_OFFSET.
3743 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3744 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3745 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3747 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3748 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3749 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3750 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3751 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3752 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3753 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3754 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3755 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3756 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3757 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3758 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3760 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3761 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3762 * config/mips/mips.c (mips_compute_frame_info): Refer to
3763 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3764 (mips_starting_frame_offset): New function.
3765 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3766 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3767 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
3768 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
3769 and return a HOST_WIDE_INT.
3770 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3771 (mmix_initial_elimination_offset): Refer to
3772 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3773 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
3774 * config/pa/pa.c (pa_starting_frame_offset): New function.
3775 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
3776 (pa_expand_prologue): Likewise.
3777 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3778 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
3779 !FRAME_GROWS_DOWNWARD handling to...
3780 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3781 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
3782 !FRAME_GROWS_DOWNWARD handling to...
3783 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3784 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
3785 !FRAME_GROWS_DOWNWARD handling to...
3786 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3787 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
3789 (rs6000_starting_frame_offset): New function.
3790 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
3791 !FRAME_GROWS_DOWNWARD handling to...
3792 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3793 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
3794 !FRAME_GROWS_DOWNWARD handling to...
3795 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3796 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
3797 !FRAME_GROWS_DOWNWARD handling to...
3798 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3799 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
3800 (rs6000_starting_frame_offset): New function.
3801 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
3802 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
3803 * config/vax/vax.c (vax_starting_frame_offset): New function.
3804 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
3805 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3806 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
3807 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
3808 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3809 * system.h (STARTING_FRAME_OFFSET): Poison.
3811 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3813 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3814 SCALAR_TYPE_MODE instead of TYPE_MODE.
3816 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3817 Alan Hayward <alan.hayward@arm.com>
3818 David Sherwood <david.sherwood@arm.com>
3820 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
3822 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3823 Alan Hayward <alan.hayward@arm.com>
3824 David Sherwood <david.sherwood@arm.com>
3826 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3828 2017-10-23 Richard Biener <rguenther@suse.de>
3830 PR tree-optimization/82129
3831 * tree-ssa-pre.c (bitmap_set_and): Remove.
3832 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3833 canonicalizing expressions in the set to those with lowest
3834 ID rather than taking that from the first edge.
3836 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3838 * combine.c (rtx_equal_for_field_assignment_p): Use
3839 byte_lowpart_offset.
3841 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3842 Alan Hayward <alan.hayward@arm.com>
3843 David Sherwood <david.sherwood@arm.com>
3845 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3846 to a SUBREG_PROMOTED_VAR.
3848 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3849 Alan Hayward <alan.hayward@arm.com>
3850 David Sherwood <david.sherwood@arm.com>
3852 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3853 (expand_debug_source_expr): Likewise.
3854 * combine.c (combine_simplify_rtx): Likewise.
3855 * cse.c (fold_rtx): Likewise.
3856 * optabs.c (expand_float): Likewise.
3857 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3858 (simplify_binary_operation_1): Likewise.
3860 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3861 Alan Hayward <alan.hayward@arm.com>
3862 David Sherwood <david.sherwood@arm.com>
3864 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3865 (record_promoted_value): Likewise.
3866 * expr.c (expand_expr_real_2): Likewise.
3867 * ree.c (update_reg_equal_equiv_notes): Likewise.
3868 (combine_set_extension): Likewise.
3869 * rtlanal.c (low_bitmask_len): Likewise.
3870 * simplify-rtx.c (neg_const_int): Likewise.
3871 (simplify_binary_operation_1): Likewise.
3873 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3874 Alan Hayward <alan.hayward@arm.com>
3875 David Sherwood <david.sherwood@arm.com>
3877 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3878 * regcprop.c (maybe_mode_change): Likewise.
3879 * reload1.c (alter_reg): Likewise.
3881 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3883 * inchash.h (inchash::hash::add_wide_int): New function.
3884 * lto-streamer-out.c (hash_tree): Use it.
3886 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3888 * inchash.h (inchash::hash::add_wide_int): Rename to...
3889 (inchash::hash::add_hwi): ...this.
3890 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3891 (polymorphic_call_target_hasher::hash): Likewise.
3892 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3893 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3894 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3895 * lto-streamer-out.c (hash_tree): Likewise.
3896 * optc-save-gen.awk: Likewise.
3897 * tree.c (add_expr): Likewise.
3899 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3902 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3903 for ordered inequality comparisons even with TARGET_IEEE_FP.
3905 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3908 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3909 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3910 Expand with cmp<dwi>_doubleword.
3912 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3914 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3915 List CET intrinsics.
3916 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3917 specific to -fcf-protection option.
3919 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3921 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3922 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3923 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3924 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3925 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3926 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3927 (extra_objs): Add cet.o for Linux/x86 targets.
3928 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3929 * config/i386/cet.c: New file.
3930 * config/i386/cetintrin.h: Likewise.
3931 * config/i386/t-cet: Likewise.
3932 * config/i386/cpuid.h (bit_SHSTK): New.
3933 (bit_IBT): Likewise.
3934 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3935 pass IBT and SHSTK bits.
3936 * config/i386/i386-builtin-types.def
3937 (VOID_FTYPE_UNSIGNED_PVOID): New.
3938 (VOID_FTYPE_UINT64_PVOID): Likewise.
3939 * config/i386/i386-builtin.def: Add CET intrinsics.
3940 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3941 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3942 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3943 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3945 * config/i386/i386.c (rest_of_insert_endbranch): New.
3946 (pass_data_insert_endbranch): Likewise.
3947 (pass_insert_endbranch): Likewise.
3948 (make_pass_insert_endbranch): Likewise.
3949 (ix86_notrack_prefixed_insn_p): Likewise.
3950 (ix86_target_string): Add -mibt, -mshstk flags.
3951 (ix86_option_override_internal): Add flag_cf_protection
3953 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3954 (ix86_print_operand): Add 'notrack' prefix output.
3955 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3956 (ix86_expand_builtin): Expand CET intrinsics.
3957 (x86_output_mi_thunk): Add 'endbranch' instruction.
3958 * config/i386/i386.h (TARGET_IBT): New.
3959 (TARGET_IBT_P): Likewise.
3960 (TARGET_SHSTK): Likewise.
3961 (TARGET_SHSTK_P): Likewise.
3962 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3963 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3964 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3965 (builtin_setjmp_setup): New pattern.
3966 (builtin_longjmp): Likewise.
3967 (rdssp<mode>): Likewise.
3968 (incssp<mode>): Likewise.
3969 (saveprevssp): Likewise.
3970 (rstorssp): Likewise.
3971 (wrss<mode>): Likewise.
3972 (wruss<mode>): Likewise.
3973 (setssbsy): Likewise.
3974 (clrssbsy): Likewise.
3975 (nop_endbr): Likewise.
3976 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3978 * config/i386/immintrin.h: Include <cetintrin.h>.
3979 * config/i386/linux-common.h
3980 (file_end_indicate_exec_stack_and_cet): New prototype.
3981 (TARGET_ASM_FILE_END): New.
3983 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3985 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3986 latencies instead of having separate table; make difference between
3987 integer and float costs.
3988 * i386.h (processor_costs): Remove scalar_stmt_cost,
3989 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3990 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3992 * x86-tune-costs.h: Remove entries which has been removed in
3993 procesor_costs from all tables; make cond_taken_branch_cost
3994 and cond_not_taken_branch_cost COST_N_INSNS based.
3996 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3998 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4000 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4002 * config/i386/i386.md (isa): Remove fma_avx512f.
4003 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4004 <avx512>_fmadd_<mode>_mask3<round_name>,
4005 <avx512>_fmsub_<mode>_mask<round_name>,
4006 <avx512>_fmsub_<mode>_mask3<round_name>,
4007 <avx512>_fnmadd_<mode>_mask<round_name>,
4008 <avx512>_fnmadd_<mode>_mask3<round_name>,
4009 <avx512>_fnmsub_<mode>_mask<round_name>,
4010 <avx512>_fnmsub_<mode>_mask3<round_name>,
4011 <avx512>_fmaddsub_<mode>_mask<round_name>,
4012 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4013 <avx512>_fmsubadd_<mode>_mask<round_name>,
4014 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4015 (*vec_widen_umult_even_v16si<mask_name>,
4016 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4017 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4019 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4021 * extend.texi: Add 'nocf_check' documentation.
4022 * gimple.texi: Add second parameter to
4023 gimple_build_call_from_tree.
4024 * invoke.texi: Add -fcf-protection documentation.
4025 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4027 2017-10-20 Richard Biener <rguenther@suse.de>
4029 PR tree-optimization/82473
4030 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4031 the largest input type.
4033 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4035 * c-attribs.c (handle_nocf_check_attribute): New function.
4036 (c_common_attribute_table): Add 'nocf_check' handling.
4037 * gimple-parser.c: Add second argument NULL to
4038 gimple_build_call_from_tree.
4039 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4040 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4042 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4043 * common.opt: Add fcf-protection flag.
4044 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4045 * flag-types.h: Add enum cf_protection_level.
4046 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4047 Add 'nocf_check' attribute propagation to gimple call.
4048 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4049 (gimple_build_call_from_tree): Update prototype.
4050 (gimple_call_nocf_check_p): New function.
4051 (gimple_call_set_nocf_check): Likewise.
4052 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4053 * ipa-icf.c: Add nocf_check attribute in statement hash.
4054 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4055 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4056 * toplev.c (process_options): Add flag_cf_protection handling.
4058 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4060 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4062 2017-10-20 Richard Biener <rguenther@suse.de>
4064 PR tree-optimization/82603
4065 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4066 remove false predicated stores.
4068 2017-10-20 Richard Biener <rguenther@suse.de>
4070 * graphite-isl-ast-to-gimple.c
4071 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4072 Remove return value and simplify, dump copied stmt after lhs
4074 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4075 Reduce dump verbosity.
4076 (gsi_insert_earliest): Likewise.
4077 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4078 * graphite.c (print_global_statistics): Adjust dumping.
4079 (print_graphite_scop_statistics): Likewise.
4080 (print_graphite_statistics): Do not dump loops here.
4081 (graphite_transform_loops): But here.
4083 2017-10-20 Nicolas Roche <roche@adacore.com>
4085 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4086 * configure: Regenerate.
4088 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4091 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4092 functions when optimizing replace GIMPLE_RETURN stmts with
4093 calls to __builtin_unreachable ().
4096 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4097 for -fsanitize=thread link of executables.
4098 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4099 link of executables.
4102 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4104 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4105 define_insns for logical vector shifts to use VI248_AVX512BW
4106 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4107 condition, useless isa and prefix attributes. Change the first
4108 2 of these define_insns to ...
4109 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4110 define_insn for avx512vl.
4111 (<shift_insn><mode>3): ... and this, new define_insn without
4112 masking for non-avx512vl.
4115 * config/i386/sse.md (*andnot<mode>3,
4116 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4117 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4118 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4119 not applied use empty suffix even for TARGET_AVX512VL.
4120 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4121 is applied, supply evex,evex or evex,evex,evex instead of just
4124 2017-10-20 Julia Koval <julia.koval@intel.com>
4126 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4127 (OPTION_MASK_ISA_GFNI_UNSET): New.
4128 (ix86_handle_option): Handle OPT_mgfni.
4129 * config/i386/cpuid.h (bit_GFNI): New.
4130 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4131 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4132 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4133 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4134 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4135 * config/i386/i386.opt: Add mgfni.
4137 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4139 * config/msp430/msp430.c (msp430_option_override): Disable
4140 -fdelete-null-pointer-checks.
4141 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4143 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4145 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4146 of x87 and SSE instructions.
4148 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4150 * asan.c (create_cond_insert_point): Do not update edge count.
4151 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4152 (afdo_propagate_circuit): Likewise.
4153 (afdo_calculate_branch_prob): Likewise.
4154 (afdo_annotate_cfg): Likewise.
4155 * basic-block.h (struct edge_def): Remove count.
4156 (edge_def::count): New accessor.
4157 * bb-reorder.c (rotate_loop): Update.
4158 (find_traces_1_round): Update.
4159 (connect_traces): Update.
4160 (sanitize_hot_paths): Update.
4161 * cfg.c (unchecked_make_edge): Update.
4162 (make_single_succ_edge): Update.
4163 (check_bb_profile): Update.
4164 (dump_edge_info): Update.
4165 (update_bb_profile_for_threading): Update.
4166 (scale_bbs_frequencies_int): Update.
4167 (scale_bbs_frequencies_gcov_type): Update.
4168 (scale_bbs_frequencies_profile_count): Update.
4169 (scale_bbs_frequencies): Update.
4170 * cfganal.c (connect_infinite_loops_to_exit): Update.
4171 * cfgbuild.c (compute_outgoing_frequencies): Update.
4172 (find_many_sub_basic_blocks): Update.
4173 * cfgcleanup.c (try_forward_edges): Update.
4174 (try_crossjump_to_edge): Update
4175 * cfgexpand.c (expand_gimple_cond): Update
4176 (expand_gimple_tailcall): Update
4177 (construct_exit_block): Update
4178 * cfghooks.c (verify_flow_info): Update
4179 (redirect_edge_succ_nodup): Update
4180 (split_edge): Update
4181 (make_forwarder_block): Update
4182 (duplicate_block): Update
4183 (account_profile_record): Update
4184 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4185 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4186 * cfgloopmanip.c (scale_loop_profile): Update.
4188 (lv_adjust_loop_entry_edge): Update.
4189 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4190 (force_nonfallthru_and_redirect): Update.
4191 (purge_dead_edges): Update.
4192 (rtl_flow_call_edges_add): Update.
4193 * cgraphunit.c (init_lowered_empty_function): Update.
4194 (cgraph_node::expand_thunk): Update.
4195 * gimple-pretty-print.c (dump_probability): Update.
4196 (dump_edge_probability): Update.
4197 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4198 * haifa-sched.c (sched_create_recovery_edges): Update.
4199 * hsa-gen.c (convert_switch_statements): Update.
4200 * ifcvt.c (dead_or_predicable): Update.
4201 * ipa-inline-transform.c (inline_transform): Update.
4202 * ipa-split.c (split_function): Update.
4203 * ipa-utils.c (ipa_merge_profiles): Update.
4204 * loop-doloop.c (add_test): Update.
4205 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4206 * lto-streamer-in.c (input_cfg): Update.
4207 (input_function): Update.
4208 * lto-streamer-out.c (output_cfg): Update.
4209 * modulo-sched.c (sms_schedule): Update.
4210 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4211 * predict.c (maybe_hot_edge_p): Update.
4212 (unlikely_executed_edge_p): Update.
4213 (probably_never_executed_edge_p): Update.
4214 (dump_prediction): Update.
4215 (drop_profile): Update.
4216 (propagate_unlikely_bbs_forward): Update.
4217 (determine_unlikely_bbs): Update.
4218 (force_edge_cold): Update.
4219 * profile.c (compute_branch_probabilities): Update.
4220 * reg-stack.c (better_edge): Update.
4221 * shrink-wrap.c (handle_simple_exit): Update.
4222 * tracer.c (better_p): Update.
4223 * trans-mem.c (expand_transaction): Update.
4224 (split_bb_make_tm_edge): Update.
4225 * tree-call-cdce.c: Update.
4226 * tree-cfg.c (gimple_find_sub_bbs): Update.
4227 (gimple_split_edge): Update.
4228 (gimple_duplicate_sese_region): Update.
4229 (gimple_duplicate_sese_tail): Update.
4230 (gimple_flow_call_edges_add): Update.
4231 (insert_cond_bb): Update.
4232 (execute_fixup_cfg): Update.
4233 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4234 * tree-complex.c (expand_complex_div_wide): Update.
4235 * tree-eh.c (lower_resx): Update.
4236 (unsplit_eh): Update.
4237 (cleanup_empty_eh_move_lp): Update.
4238 * tree-inline.c (copy_edges_for_bb): Update.
4239 (freqs_to_counts): Update.
4240 (copy_cfg_body): Update.
4241 * tree-ssa-dce.c (remove_dead_stmt): Update.
4242 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4243 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4244 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4245 (unloop_loops): Update.
4246 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4247 * tree-ssa-loop-split.c (connect_loops): Update.
4248 (split_loop): Update.
4249 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4250 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4251 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4252 * tree-ssa-reassoc.c (branch_fixup): Update.
4253 * tree-ssa-tail-merge.c (replace_block_by): Update.
4254 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4255 (compute_path_counts): Update.
4256 (update_profile): Update.
4257 (recompute_probabilities): Update.
4258 (update_joiner_offpath_counts): Update.
4259 (estimated_freqs_path): Update.
4260 (freqs_to_counts_path): Update.
4261 (clear_counts_path): Update.
4262 (ssa_fix_duplicate_block_edges): Update.
4263 (duplicate_thread_path): Update.
4264 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4265 (case_bit_test_cmp): Update.
4266 (collect_switch_conv_info): Update.
4267 (gen_inbound_check): Update.
4268 (do_jump_if_equal): Update.
4269 (emit_cmp_and_jump_insns): Update.
4270 * tree-tailcall.c (decrease_profile): Update.
4271 (eliminate_tail_call): Update.
4272 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4273 (vect_do_peeling): Update.
4274 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4275 * ubsan.c (ubsan_expand_null_ifn): Update.
4276 (ubsan_expand_ptr_ifn): Update.
4277 * value-prof.c (gimple_divmod_fixed_value): Update.
4278 (gimple_mod_pow2): Update.
4279 (gimple_mod_subtract): Update.
4280 (gimple_ic): Update.
4281 (gimple_stringop_fixed_value): Update.
4283 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4286 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4288 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4290 PR rtl-optimization/82395
4291 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4292 based on non_spilled_static_chain_regno_p.
4294 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4296 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4297 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4298 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4300 2017-10-19 Martin Sebor <msebor@redhat.com>
4302 PR tree-optimization/82596
4303 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4305 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4307 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4308 (get_mem_refs_of_builtin_call): Likewise.
4309 * builtins.c (expand_builtin_apply): Adjust call to
4310 allocate_dynamic_stack_space.
4311 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4312 the third argument to allocate_dynamic_stack_space, otherwise -1.
4313 (expand_builtin): Deal with all alloca variants.
4314 (is_inexpensive_builtin): Likewise.
4315 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4316 * calls.c (special_function_p): Deal with all alloca variants.
4317 (initialize_argument_information): Adjust call to
4318 allocate_dynamic_stack_space.
4319 (expand_call): Likewise.
4320 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4321 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4322 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4323 use it for the stack usage computation.
4324 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4325 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4326 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4327 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4328 (in_loop_p): Remove first argument and useless check.
4329 (pass_walloca::execute): Remove useless test and adjust call to above.
4330 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4331 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4332 (gimplify_call_expr): Deal with all alloca variants.
4333 * hsa-gen.c (gen_hsa_alloca): Likewise.
4334 (gen_hsa_insns_for_call): Likewise.
4335 * ipa-pure-const.c (special_builtin_state): Likewise.
4336 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4337 * tree-object-size.c (alloc_object_size): Likewise.
4338 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4339 (call_may_clobber_ref_p_1): Likewise.
4340 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4341 (ccp_fold_stmt): Likewise.
4342 (optimize_stack_restore): Likewise.
4343 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4344 (mark_all_reaching_defs_necessary_1): Likewise.
4345 (propagate_necessity): Likewise.
4346 (eliminate_unnecessary_stmts): Likewise.
4347 * tree.c (build_common_builtin_nodes): Build
4348 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4349 (build_alloca_call_expr): New function.
4350 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4351 (CASE_BUILT_IN_ALLOCA): Likewise.
4352 (build_alloca_call_expr): Declare.
4353 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4355 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4358 * dwarf2out.c (new_die_raw): New static inline function.
4359 (new_die): Use it to create the DIE.
4360 (add_AT_external_die_ref): Likewise.
4361 (clone_die): Likewise.
4362 (clone_as_declaration): Likewise.
4363 (dwarf2out_vms_debug_main_pointer): Likewise.
4364 (base_type_die): Likewise. Remove early return for corner cases.
4365 Do not call add_pubtype on the DIE here.
4366 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4367 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4368 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4369 native order exists for base types, attach the DIE manually and call
4370 add_pubtype on it. Do not equate a reverse order DIE to the type.
4372 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4374 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4375 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4377 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4379 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4381 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4382 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4383 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4384 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4385 * opts.c (sanitizer_opts): Add builtin.
4386 * ubsan.c (instrument_builtin): New function.
4387 (pass_ubsan::execute): Call it.
4388 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4389 * doc/invoke.texi: Document -fsanitize=builtin.
4391 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4392 builtins, store max (log2 (align), 0) into uchar field instead of
4393 align into uptr field.
4394 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4395 store uchar 0 field instead of uptr 0 field.
4396 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4397 instead of passing one address of struct with 2 locations pass
4398 two addresses of structs with 1 location each.
4399 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4400 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4401 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4402 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4403 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4404 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4405 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4406 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4408 2017-10-19 Martin Liska <mliska@suse.cz>
4411 * file-find.c (remove_prefix): Remove.
4412 * file-find.h (remove_prefix): Likewise.
4413 * gcc-ar.c: Remove smartness of lookup.
4415 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4417 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4418 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4419 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4421 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4424 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4425 (setcc + and to xor + setcc): New peephole2.
4427 2017-10-19 Tom de Vries <tom@codesourcery.com>
4429 * doc/sourcebuild.texi (Test Directives, Variants of
4430 dg-require-support): Add dg-require-stack-size.
4432 2017-10-19 Martin Liska <mliska@suse.cz>
4435 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4436 that have a large alignment.
4437 (gimplify_target_expr): Likewise.
4439 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4441 PR rtl-optimization/82602
4442 * ira.c (rtx_moveable_p): Return false for volatile asm.
4444 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4447 * config/i386/i386-modes.def (CCGZ): New CC mode.
4448 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4449 * config/i386/predicates.md (ix86_comparison_operator):
4451 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4452 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4453 with double-word subtraction.
4454 (put_condition_code): Handle CCGZmode.
4456 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4458 * wide-int.cc (debug (const wide_int &)): New.
4459 (debug (const wide_int *)): New.
4460 (debug (const widest_int &)): New.
4461 (debug (const widest_int *)): New.
4463 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4466 * lra-constraints.c (curr_insn_transform): Use non-input operand
4467 instead of output one for matched reload.
4469 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4471 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4472 (tree-ssa-loop-ivopts.h): New header file.
4473 (struct builtin_info): New fields.
4474 (classify_builtin_1): Compute and record base and offset parts for
4475 memset builtin partition by calling strip_offset.
4476 (offset_cmp, fuse_memset_builtins): New functions.
4477 (finalize_partitions): Fuse adjacent memset partitions by calling
4479 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4480 Expose the interface.
4481 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4483 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4485 PR tree-optimization/82574
4486 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4487 that data reference must be executed exactly once per iteration
4488 against the outermost loop in nest.
4489 (classify_partition): Update call to above function.
4491 2017-10-18 Richard Biener <rguenther@suse.de>
4493 PR tree-optimization/82591
4494 * graphite.c (graphite_transform_loops): Move code gen message
4496 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4497 Here. Handle scop_to_isl_ast failing.
4498 (scop_to_isl_ast): Limit the number of ISL operations.
4500 2017-10-18 Richard Biener <rguenther@suse.de>
4502 * graphite-isl-ast-to-gimple.c
4503 (translate_isl_ast_to_gimple::set_rename): Simplify.
4504 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4505 (graphite_copy_stmts_from_block): ... here.
4506 (copy_bb_and_scalar_dependences): Simplify.
4507 (add_parameters_to_ivs_params): Canonicalize.
4508 (generate_entry_out_of_ssa_copies): Simplify.
4509 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4510 by passing in ISL dimension.
4511 (parameter_index_in_region_1): Rename to ...
4512 (parameter_index_in_region): ... this.
4513 (extract_affine): Adjust assert, pass down parameter index.
4514 (add_param_constraints): Use range-info when available.
4515 (build_scop_context): Adjust.
4516 * sese.c (new_sese_info): Adjust.
4517 (free_sese_info): Likewise.
4518 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4519 Remove unused typedefs.
4520 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4522 2017-10-18 Martin Liska <mliska@suse.cz>
4524 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4526 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4528 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4529 (s390_sched_init): Do not reset s390_sched_state if we entered the
4530 current basic block via a fallthru edge and all others are unlikely.
4532 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4534 * config/s390/s390.c (NUM_SIDES): New variable.
4535 (LONGRUNNING_THRESHOLD): New variable.
4536 (LATENCY_FACTOR): New variable.
4537 (s390_sched_score): Decrease score for long-running instructions on
4539 (s390_sched_variable_issue): Perform bookkeeping for long-running
4542 2017-10-18 Richard Biener <rguenther@suse.de>
4544 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4545 Simplify with removal of the parameter rename map.
4546 (set_rename): Likewise.
4547 (should_copy_to_new_region): Likewise.
4548 (graphite_copy_stmts_from_block): Likewise.
4549 (copy_bb_and_scalar_dependences): Remove initialization of
4550 unused copied_bb_map.
4552 (copy_internal_parameters): Likewise.
4553 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4554 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4555 Use INTEGRAL_TYPE_P.
4556 (parameter_index_in_region_1): Rename to ...
4557 (assign_parameter_index_in_region): ... this. Assert we have
4558 a parameter we handle.
4559 (scan_tree_for_params): Adjust.
4560 * sese.h (parameter_rename_map_t): Remove.
4561 (struct sese_info_t): Remove unused parameter_rename_map and
4562 copied_bb_map members.
4563 * sese.c (new_sese_info): Adjust.
4564 (free_sese_info): Likewise.
4566 2017-10-18 Martin Liska <mliska@suse.cz>
4569 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
4570 on an abnormal edge.
4572 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
4574 * doc/invoke.texi (ffunction-sections and fdata-sections):
4577 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4579 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
4580 the use statement can throw internally.
4582 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4584 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4585 any RTX present on the RHS of a SET.
4586 * compare-elim.c (try_eliminate_compare): Restore comment.
4588 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4590 * langhooks.h (struct lang_hooks): Document that tree_size langhook
4591 may be also called on tcc_type nodes.
4592 * langhooks.c (lhd_tree_size): Likewise.
4594 2017-10-17 David Malcolm <dmalcolm@redhat.com>
4596 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
4597 format_warning_at_substring.
4598 (maybe_warn): Convert source_range * param to a location_t. Pass
4599 UNKNOWN_LOCATION rather than NULL to fmtwarn.
4600 (format_directive): Remove code to extract source_ranges and
4601 source_range * in favor of just a location_t.
4602 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
4604 * substring-locations.c (format_warning_va): Convert
4605 source_range * param to a location_t.
4606 (format_warning_at_substring): Likewise.
4607 * substring-locations.h (format_warning_va): Likewise.
4608 (format_warning_at_substring): Likewise.
4610 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
4612 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
4614 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
4615 and scatter/gather ops.
4617 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
4618 vec_gather_load and vec_scatter_store.
4619 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
4620 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
4621 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
4622 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
4623 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
4624 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
4626 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
4628 * reg-stack.c (compare_for_stack_reg): Add bool argument.
4629 Detect FTST instruction and handle its register pops. Only pop
4630 second operand if can_pop_second_op is true.
4631 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
4632 set can_pop_second_op to false in the compare_for_stack_reg call.
4634 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
4635 output_fp_compare for stack register operands.
4636 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
4637 instructions here. Do not emit stack register pops here. Assert
4638 that FCOMPP pops next to top stack register. Rewrite function.
4640 2017-10-17 Nathan Sidwell <nathan@acm.org>
4643 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
4644 use DECL_ASSEMBLER_NAME_RAW.
4647 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
4650 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
4651 Wilco Dijkstra <wilco.dijkstra@arm.com>
4653 * builtins.c (expand_builtin_update_setjmp_buf): Add a
4654 converstion to Pmode from the buf_addr.
4656 2017-10-17 Richard Biener <rguenther@suse.de>
4658 * graphite-dependences.c (scop_get_reads_and_writes): Change
4659 output parameters to references.
4661 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
4663 PR 71026/tree-optimization
4664 * fold-const.c (distribute_real_division): Removed.
4665 (fold_binary_loc): Remove calls to distribute_real_divison.
4667 2017-10-17 Richard Biener <rguenther@suse.de>
4669 * graphite-scop-detection.c
4670 (scop_detection::stmt_has_simple_data_refs_p): Always use
4671 the full nest as region.
4672 (try_generate_gimple_bb): Likewise.
4673 * sese.c (scalar_evolution_in_region): Simplify now that
4674 SCEV can handle instantiation in regions.
4675 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
4676 in the non-loop part of a function if requested.
4678 2017-10-17 Richard Biener <rguenther@suse.de>
4680 PR tree-optimization/82563
4681 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
4683 (graphite_regenerate_ast_isl): Call it.
4684 * graphite-scop-detection.c (build_scops): Remove entry edge split.
4686 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4688 PR tree-optimization/82549
4689 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
4690 Formatting fixes. Instead of calling make_bit_field_ref with negative
4693 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
4695 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
4696 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
4697 _mm_maskz_reduce_ss): New.
4698 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
4699 __builtin_ia32_reducess_mask): Ditto..
4700 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
4701 * config/i386/sse.md (reduces<mode>): Renamed to ...
4702 (reduces<mode><mask_scalar_name>): ... this.
4703 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
4705 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
4706 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
4708 2017-10-16 David Malcolm <dmalcolm@redhat.com>
4710 * Makefile.in (OBJS): Add unique-ptr-tests.o.
4711 * selftest-run-tests.c (selftest::run_tests): Call
4712 selftest::unique_ptr_tests_cc_tests.
4713 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
4714 * unique-ptr-tests.cc: New file.
4716 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
4719 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4721 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4722 (make_hard_regno_born, make_hard_regno_dead): Update
4723 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
4725 2017-10-16 Jeff Law <law@redhat.com>
4727 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4729 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
4731 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
4733 2017-10-16 Olivier Hainque <hainque@adacore.com>
4735 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
4736 with_cpu if we were configured for an e500v2 target cpu name.
4738 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4740 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4741 * doc/invoke.texi: Document +nodsp as a valid extension for
4744 2017-10-16 Martin Liska <mliska@suse.cz>
4746 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4747 (test_set_range): Likewise.
4748 (test_range_functions): Rename to ...
4749 (test_bit_in_range): ... this.
4750 (sbitmap_c_tests): Add new test.
4752 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4754 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4756 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4757 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4759 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4761 * config/aarch64/aarch64-builtins.c
4762 (aarch64_types_quadopu_lane_qualifiers): New.
4763 (TYPES_QUADOPU_LANE): New.
4764 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4765 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4766 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4767 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
4768 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
4769 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
4770 (Vdottype, DOTPROD): New.
4771 (sur): Add SDOT and UDOT.
4773 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4775 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4776 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
4777 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
4779 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
4780 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
4781 Enable TARGET_DOTPROD.
4782 (cortex-a75.cortex-a55): Likewise.
4783 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
4785 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4787 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
4788 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
4790 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
4792 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
4793 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
4794 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
4795 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
4796 * config/arm/types.md (neon_dot, neon_dot_q): New.
4797 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
4799 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4801 * config/arm/arm.h (TARGET_DOTPROD): New.
4802 * config/arm/arm.c (arm_arch_dotprod): New.
4803 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
4804 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
4805 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
4806 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
4807 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
4808 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
4809 * doc/invoke.texi (armv8.2-a): Document dotprod
4811 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4813 * i386.c (ix86_vec_cost): New function.
4814 (ix86_rtx_costs): Handle vector operations better.
4815 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
4816 * x86-tune-costs.h: Add new costs to all tables.
4818 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4820 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4822 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4823 divsd, sqrtss and sqrtsd
4824 * x86-tune-costs.h: Add new entries to all costs.
4825 (znver1_cost): Fix to match real instruction latencies.
4827 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4828 Michael Collison <michael.collison@arm.com>
4830 * compare-elim.c: Include emit-rtl.h.
4831 (can_merge_compare_into_arith): New function.
4832 (try_validate_parallel): Likewise.
4833 (try_merge_compare): Likewise.
4834 (try_eliminate_compare): Call the above when no previous clobber
4836 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4839 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4843 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4844 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4848 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4849 to be any operand_equal_p operands. For & (B - 1) require
4850 B to be power of 2. Recognize
4851 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4853 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4856 * optabs.c (expand_memory_blockage): Fix call of
4857 targetm.have_memory_blockage.
4859 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4862 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4863 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4864 objects to extra_objs instead of overwriting it.
4866 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4868 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4869 Use any_fp_register_operand as operand[3] predicate. Simplify
4870 equality test for operands[2] and operands[4] memory location.
4871 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4872 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4873 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4874 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4875 any_fp_register_operand as operand[1] predicate. Simplify
4876 equality test for operands[0] and operands[3] memory location.
4877 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4878 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4879 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4881 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4883 * target-insns.def: Add memory_blockage.
4884 * optabs.c (expand_memory_blockage): New function.
4885 (expand_asm_memory_barrier): Rename ...
4886 (expand_asm_memory_blockage): ... to this.
4887 (expand_mem_thread_fence): Call expand_memory_blockage
4888 instead of expand_asm_memory_barrier.
4889 (expand_mem_singnal_fence): Ditto.
4890 (expand_atomic_load): Ditto.
4891 (expand_atomic_store): Ditto.
4892 * doc/md.texi (Standard Pattern Names For Generation):
4893 Document memory_blockage instruction pattern.
4895 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4897 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4898 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4899 * config/rl78/rl78.md: New define_expand "adddi3".
4901 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4903 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4906 2017-10-13 Jeff Law <law@redhat.com>
4908 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4910 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4913 * internal-fn.c (expand_mul_overflow): If both operands have
4914 the same highpart of -1 or 0 and the topmost bit of lowpart
4915 is different, overflow is if res <= 0 rather than res < 0.
4917 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4919 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4920 TARGET_P9_VECTOR code for unaligned_load case.
4922 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4924 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4926 2017-10-13 Nathan Sidwell <nathan@acm.org>
4928 * tree-core.h (tree_contains_struct): Make bool.
4929 * tree.c (tree_contains_struct): Likewise.
4930 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4931 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4932 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4933 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4934 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4936 2017-10-13 Richard Biener <rguenther@suse.de>
4938 * graphite-isl-ast-to-gimple.c
4939 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4940 parameters and dominance check.
4941 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4942 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4943 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4944 Do not update SSA form here or do intermediate IL verification.
4945 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4946 (graphite_initialize): Remove check on the number of loops in
4947 the function and inline into graphite_transform_loops.
4948 (graphite_finalize): Inline into graphite_transform_loops.
4949 (graphite_transform_loops): Perform SSA update and IL verification
4951 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4953 2017-10-13 Richard Biener <rguenther@suse.de>
4955 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4956 graphite_expression_type_precision): Avoid global constructor
4958 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4959 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4960 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4961 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4962 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4963 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4965 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4968 * config/i386/i386.h (ix86_red_zone_size): New.
4969 * config/i386/i386.md (push peephole2s): Replace
4970 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4972 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4973 Alan Hayward <alan.hayward@arm.com>
4974 David Sherwood <david.sherwood@arm.com>
4976 * combine.c (can_change_dest_mode): Reject changes in
4977 REGMODE_NATURAL_SIZE.
4979 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4980 Alan Hayward <alan.hayward@arm.com>
4981 David Sherwood <david.sherwood@arm.com>
4983 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4984 (expand_debug_source_expr): Likewise.
4985 * combine.c (combine_simplify_rtx): Likewise.
4986 * cse.c (fold_rtx): Likewise.
4987 * fwprop.c (canonicalize_address): Likewise.
4988 * targhooks.c (default_shift_truncation_mask): Likewise.
4990 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4991 Alan Hayward <alan.hayward@arm.com>
4992 David Sherwood <david.sherwood@arm.com>
4994 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4995 (widened_mode): Likewise.
4996 (expand_unop): Likewise.
4997 * ree.c (transform_ifelse): Likewise.
4998 (merge_def_and_ext): Likewise.
4999 (combine_reaching_defs): Likewise.
5000 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5002 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5003 Alan Hayward <alan.hayward@arm.com>
5004 David Sherwood <david.sherwood@arm.com>
5006 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5007 * combine.c (gen_lowpart_for_combine): Likewise.
5008 * dwarf2out.c (rtl_for_decl_location): Likewise.
5009 * final.c (alter_subreg): Likewise.
5010 * rtlhooks.c (gen_lowpart_general): Likewise.
5011 (gen_lowpart_if_possible): Likewise.
5013 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5014 Alan Hayward <alan.hayward@arm.com>
5015 David Sherwood <david.sherwood@arm.com>
5017 * calls.c (expand_call): Use subreg_lowpart_offset.
5018 * cse.c (cse_insn): Likewise.
5019 * regcprop.c (copy_value): Likewise.
5020 (copyprop_hardreg_forward_1): Likewise.
5022 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5025 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5026 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5027 =Q constraints to +Q and into insn condition add check
5028 that operands[0] and operands[1] are equal.
5029 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5030 =Q constraints to +Q and into insn condition add check
5031 that operands[0] is equal to either operands[1] or operands[2].
5034 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5035 instead of handling MINUS_EXPR twice (once for each argument),
5036 canonicalize operand order and handle just once, use rtype where
5037 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5040 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5041 any values of __C while still being pattern recognizable as a simple
5044 2017-10-13 Richard Biener <rguenther@suse.de>
5046 PR tree-optimization/82451
5048 2017-10-02 Richard Biener <rguenther@suse.de>
5050 PR tree-optimization/82355
5051 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5052 a mapping for the enclosing loop but avoid generating one for
5054 (copy_bb_and_scalar_dependences): Remove premature codegen
5055 error on PHIs in blocks duplicated into multiple places.
5056 * graphite-scop-detection.c
5057 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5058 in the region use it as loop and nest to analyze the DR in.
5059 (try_generate_gimple_bb): Likewise.
5060 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5061 (add_loop_constraints): For blocks in a loop not in the region
5062 create a dimension with a single iteration.
5063 * sese.h (gbb_loop_at_index): Remove assert.
5065 * cfgloop.c (loop_preheader_edge): For the loop tree root
5066 return the single successor of the entry block.
5067 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5068 Reset the SCEV hashtable and niters.
5069 * graphite-scop-detection.c
5070 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5071 assert that we only have POLYNOMIAL_CHREC that vary in loops
5072 contained in the region.
5073 (scop_detection::graphite_can_represent_expr): Adjust.
5074 (scop_detection::stmt_has_simple_data_refs_p): For loops
5075 not in the region set loop to NULL. The nest is now the
5076 entry edge to the region.
5077 (try_generate_gimple_bb): Likewise.
5078 * sese.c (scalar_evolution_in_region): Adjust for
5079 instantiate_scev change.
5080 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5081 Make nest parameter the edge into the region.
5082 (create_data_ref): Likewise.
5083 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5084 entry edge into a region and adjust instantiate_scev calls.
5085 (create_data_ref): Likewise.
5086 (graphite_find_data_references_in_stmt): Likewise.
5087 (find_data_references_in_stmt): Pass the loop preheader edge
5088 from the nest argument.
5089 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5090 parameter the edge into the region.
5091 (instantiate_parameters): Use the loop preheader edge as entry.
5092 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5094 (get_instantiated_value_entry): Make instantiate_below parameter
5095 the edge into the region.
5096 (instantiate_scev_name): Likewise. Adjust dominance checks,
5097 when we cannot use loop-based instantiation instantiate by
5098 walking use-def chains.
5099 (instantiate_scev_poly): Adjust.
5100 (instantiate_scev_binary): Likewise.
5101 (instantiate_scev_convert): Likewise.
5102 (instantiate_scev_not): Likewise.
5103 (instantiate_array_ref): Remove.
5104 (instantiate_scev_3): Likewise.
5105 (instantiate_scev_2): Likewise.
5106 (instantiate_scev_1): Likewise.
5107 (instantiate_scev_r): Do not blindly handle N-operand trees.
5108 Do not instantiate array-refs. Handle all constants and invariants.
5109 (instantiate_scev): Make instantiate_below parameter
5110 the edge into the region.
5111 (resolve_mixers): Use the loop preheader edge for the region
5112 parameter to instantiate_scev_r.
5113 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5115 2017-10-13 Richard Biener <rguenther@suse.de>
5117 PR tree-optimization/82525
5118 * graphite-isl-ast-to-gimple.c
5119 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5121 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5122 Fail code generation when we cannot represent the isl integer.
5123 (binary_op_to_tree): Elide modulo operations that are no-ops
5124 in the type we code generate. Remove now superfluous code
5127 2017-10-13 Richard Biener <rguenther@suse.de>
5129 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5130 (scop_detection::harmful_loop_in_region): Remove premature
5131 IV type restriction.
5132 (scop_detection::graphite_can_represent_scev): We can handle
5133 pointer IVs just fine.
5135 2017-10-13 Alan Modra <amodra@gmail.com>
5137 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5138 "Clobbers and Scratch Registers". Add paragraph on
5139 alternative to clobbers for scratch registers and OpenBLAS
5142 2017-10-13 Alan Modra <amodra@gmail.com>
5144 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5145 example of a memory input for a string of known length. Move
5146 commentary out of table. Add a number of new examples
5147 covering array memory inputs.
5149 2017-10-12 Martin Liska <mliska@suse.cz>
5151 PR tree-optimization/82493
5152 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5153 (test_range_functions): New function.
5154 (sbitmap_c_tests): Likewise.
5155 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5156 * selftest.h (sbitmap_c_tests): New function.
5158 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5160 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5162 * config/rs6000/amo.h: Fix spacing issue.
5164 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5167 * config/i386/i386.md (*ashl<mode>3_mask_1,
5168 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5169 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5172 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5174 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5175 (profile_probability): Set max_probability
5176 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5178 * profile-count.c (profile_probability::differs_from_p): Do not
5179 rely on max_probaiblity == 10000
5181 2017-10-12 Jeff Law <law@redhat.com>
5183 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5186 2017-10-12 Martin Sebor <msebor@redhat.com>
5190 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5191 (handle_alias_pairs): Call it.
5192 * common.opt (-Wattribute-alias): New option.
5193 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5194 * doc/invoke.texi (-Wattribute-alias): Document.
5196 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5199 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5201 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5203 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5204 (make_hard_regno_born, make_hard_regno_dead): Update
5205 bb_killed_pseudos and bb_gen_pseudos.
5207 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5209 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5211 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5213 * config/alpha/alpha.c (alpha_split_conditional_move):
5214 Use std::swap instead of manually swapping.
5215 (alpha_stdarg_optimize_hook): Ditto.
5216 (alpha_canonicalize_comparison): Ditto.
5218 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5220 * tree-loop-distribution.c (struct builtin_info): New struct.
5221 (struct partition): Refactor fields into struct builtin_info.
5222 (partition_free): Free struct builtin_info.
5223 (build_size_arg_loc, build_addr_arg_loc): Delete.
5224 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5225 information from struct builtin_info.
5226 (find_single_drs): New function refactored from classify_partition.
5227 Also moved builtin validity checks to this function.
5228 (compute_access_range, alloc_builtin): New functions.
5229 (classify_builtin_st, classify_builtin_ldst): New functions.
5230 (classify_partition): Refactor code into functions find_single_drs,
5231 classify_builtin_st and classify_builtin_ldst.
5232 (distribute_loop): Don't do runtime alias check when distributing
5234 (find_seed_stmts_for_distribution): New function.
5235 (pass_loop_distribution::execute): Refactor code finding seed
5236 stmts into above function. Support distribution for the innermost
5237 two-level loop nest. Adjust dump information.
5239 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5241 * tree-loop-distribution.c: Adjust the general comment.
5242 (NUM_PARTITION_THRESHOLD): New macro.
5243 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5244 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5245 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5246 in call to build_partition_graph.
5247 (finalize_partitions): New parameter. Make loop distribution more
5248 conservative by fusing more partitions.
5249 (distribute_loop): Don't do runtime alias check in case of loop nest
5251 (find_seed_stmts_for_distribution): New function.
5252 (prepare_perfect_loop_nest): New function.
5253 (pass_loop_distribution::execute): Refactor code finding seed stmts
5254 and loop nest into above functions. Support loop nest distribution.
5255 Adjust dump information accordingly.
5257 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5259 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5260 and set PTYPE_SEQUENTIAL for merged partition.
5262 2017-10-12 Richard Biener <rguenther@suse.de>
5264 PR tree-optimization/69728
5266 2017-09-19 Richard Biener <rguenther@suse.de>
5268 PR tree-optimization/69728
5269 * graphite-sese-to-poly.c (schedule_error): New global.
5270 (add_loop_schedule): Handle empty domain by failing the
5272 (build_original_schedule): Handle schedule_error.
5274 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5275 domain by returning an unchanged schedule.
5277 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5279 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5280 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5282 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5284 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5287 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5290 * expr.c (store_field): Don't optimize away bitsize == 0 store
5291 from CALL_EXPR with addressable return type.
5293 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5295 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5296 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5297 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5298 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5299 TARGET_ISEL instead of TARGET_ISEL<sel>.
5301 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5303 * config/rs6000/rs6000.c
5304 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5306 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5308 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5310 (reg_or_cint_operand): Fix comment.
5311 (reg_or_zero_operand): New predicate.
5312 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5313 * config/rs6000/rs6000.c (output_isel): Delete.
5314 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5315 instead of reg_or_cint_operand. Output instruction directly (not via
5317 (isel_unsigned_<mode>): Ditto.
5318 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5319 gpc_reg_operand. Add an instruction alternative for this. Output
5320 instruction directly.
5321 (*isel_reversed_unsigned_<mode>): Ditto.
5323 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5325 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5326 (TARGET_CANONICALIZE_COMPARISON): Define.
5328 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5331 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5332 Check whether the dest is REG before adding REG_EQUIV note.
5334 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5337 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5339 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5340 (make_hard_regno_born, make_hard_regno_dead): Update
5341 bb_killed_pseudos and bb_gen_pseudos.
5343 2017-10-11 Nathan Sidwell <nathan@acm.org>
5345 * incpath.h (enum incpath_kind): Name enum, prefix values.
5346 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5347 * incpath.c (heads, tails): Use INC_MAX.
5348 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5349 (merge_include_chains, split_quote_chain,
5350 register_include_chains): Update incpath_kind names.
5351 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5352 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5354 (add_framework_path, darwin_register_objc_includes): Likewise.
5355 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5357 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5359 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5360 Do not use float_operator operator predicate.
5361 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5362 * config/i386/predicates.md (float_operator): Remove predicate.
5364 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5366 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5367 (*jccxf_i387): Ditto.
5368 (*jcc<mode>_i387): Ditto.
5369 (*jccu<mode>_i387): Ditto.
5370 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5371 (*jcc_*_i387 splitters): Remove.
5372 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5373 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5374 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5377 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5379 * profile-count.h (slow_safe_scale_64bit): New function.
5380 (safe_scale_64bit): New inline.
5381 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5382 * profile-count.c: Include wide-int.h
5383 (slow_safe_scale_64bit): New.
5385 2017-10-11 Nathan Sidwell <nathan@acm.org>
5387 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5388 HAS_DECL_ASSEMBLER_NAME_P.
5389 * gimple-expr.c (gimple_decl_printable_name: Check
5390 HAS_DECL_ASSEMBLER_NAME_P too.
5391 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5392 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5393 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5394 * passes.c (rest_of_decl_compilation): Check
5395 HAS_DECL_ASSEMBLER_NAME_P too.
5396 * recog.c (verify_changes): Likewise.
5397 * tree-pretty-print.c (dump_decl_name): Likewise.
5398 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5400 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5401 (SET_DECL_ASSEMBLER_NAME): Use it.
5402 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5403 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5404 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5406 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5408 * config.gcc (i386, x86_64): Add extra objects.
5409 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5410 (ix86_min_insn_size): Declare.
5411 (ix86_issue_rate): Declare.
5412 (ix86_adjust_cost): Declare.
5413 (ia32_multipass_dfa_lookahead): Declare.
5414 (ix86_macro_fusion_p): Declare.
5415 (ix86_macro_fusion_pair_p): Declare.
5416 (ix86_bd_has_dispatch): Declare.
5417 (ix86_bd_do_dispatch): Declare.
5418 (ix86_core2i7_init_hooks): Declare.
5419 (ix86_atom_sched_reorder): Declare.
5420 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5421 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5422 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5423 (rip_relative_addr_p): Rename to ...
5424 (ix86_rip_relative_addr_p): ... this one; export.
5425 (memory_address_length): Update.
5426 (ix86_issue_rate): Move to x86-tune-sched.c.
5427 (ix86_flags_dependent): Move to x86-tune-sched.c.
5428 (ix86_agi_dependent): Move to x86-tune-sched.c.
5429 (exact_dependency_1): Move to x86-tune-sched.c.
5430 (exact_store_load_dependency): Move to x86-tune-sched.c.
5431 (ix86_adjust_cost): Move to x86-tune-sched.c.
5432 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5433 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5434 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5435 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5436 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5437 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5438 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5439 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5440 (min_insn_size): Rename to ...
5441 (ix86_min_insn_size): ... this one; export.
5442 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5443 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5444 (core2i7_first_cycle_multipass_backtrack): Move to
5445 x86-tune-sched-core.c.
5446 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5447 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5448 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5449 (ix86_avoid_jump_mispredicts): Update.
5450 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5451 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5452 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5453 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5454 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5455 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5456 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5457 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5458 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5459 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5460 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5461 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5462 (BIG): Move to ix86-tune-sched-bd.c.
5463 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5464 (enum insn_path): Move to ix86-tune-sched-bd.c.
5465 (get_mem_group): Move to ix86-tune-sched-bd.c.
5466 (is_cmp): Move to ix86-tune-sched-bd.c.
5467 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5468 (is_branch): Move to ix86-tune-sched-bd.c.
5469 (is_prefetch): Move to ix86-tune-sched-bd.c.
5470 (init_window): Move to ix86-tune-sched-bd.c.
5471 (allocate_window): Move to ix86-tune-sched-bd.c.
5472 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5473 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5474 (process_end_window): Move to ix86-tune-sched-bd.c.
5475 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5476 (find_constant): Move to ix86-tune-sched-bd.c.
5477 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5478 (has_immediate): Move to ix86-tune-sched-bd.c.
5479 (get_insn_path): Move to ix86-tune-sched-bd.c.
5480 (get_insn_group): Move to ix86-tune-sched-bd.c.
5481 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5482 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5483 (add_insn_window): Move to ix86-tune-sched-bd.c.
5484 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5485 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5486 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5487 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5488 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5489 (do_dispatch): Move to ix86-tune-sched-bd.c.
5490 (has_dispatch): Move to ix86-tune-sched-bd.c.
5491 * config/i386/t-i386: Add new object files.
5492 * config/i386/x86-tune-costs.h: New file.
5493 * config/i386/x86-tune-sched-atom.c: New file.
5494 * config/i386/x86-tune-sched-bd.c: New file.
5495 * config/i386/x86-tune-sched-core.c: New file.
5496 * config/i386/x86-tune-sched.c: New file.
5498 2017-10-11 Liu Hao <lh_mouse@126.com>
5500 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5501 the generic version below instead.
5502 (should_colorize): Recognize Windows consoles as terminals
5504 * pretty-print.c [__MINGW32__] (write_all): New function.
5505 [__MINGW32__] (find_esc_head): Likewise.
5506 [__MINGW32__] (find_esc_terminator): Likewise.
5507 [__MINGW32__] (eat_esc_sequence): Likewise.
5508 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5510 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5513 2017-10-11 Richard Biener <rguenther@suse.de>
5515 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5516 Properly call analyze_scalar_evolution with the loop of the stmt.
5518 2017-10-11 Richard Biener <rguenther@suse.de>
5520 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5521 * tree-core.h (tree_base): Add chrec_var union member.
5522 * tree.h (CHREC_VAR): Remove.
5523 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5524 * tree-chrec.h (build_polynomial_chrec): Adjust.
5525 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5526 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5528 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5530 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5531 * match.pd: ... here.
5532 ((T) X == (T) Y): Relax condition.
5534 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5536 PR tree-optimization/82472
5537 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5539 (break_alias_scc_partitions): Update postorder number.
5541 2017-10-11 Martin Liska <mliska@suse.cz>
5544 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5546 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5548 2017-10-11 Martin Liska <mliska@suse.cz>
5550 * print-rtl.c (print_insn): Move declaration of idbuf
5551 to same scope as name.
5553 2017-10-11 Martin Liska <mliska@suse.cz>
5558 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5561 2017-10-11 Martin Liska <mliska@suse.cz>
5564 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5567 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
5569 * config.gcc (default_gnu_indirect_function): Default to yes for
5570 arm*-*-linux* with glibc.
5572 2017-10-11 Richard Biener <rguenther@suse.de>
5574 * tree-scalar-evolution.c (get_scalar_evolution): Handle
5575 default-defs and types we do not want to analyze.
5576 (interpret_loop_phi): Replace unreachable code with an assert.
5577 (compute_scalar_evolution_in_loop): Remove and inline ...
5578 (analyze_scalar_evolution_1): ... here, replacing condition with
5579 what makes the intent clearer. Remove handling of cases
5580 get_scalar_evolution now handles.
5582 2017-10-10 Jim Wilson <wilson@tuliptree.org>
5584 PR rtl-optimization/81434
5585 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5586 comment for main loop. In sched_group_found if, also add checks for
5587 pass and min_cost_group.
5589 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
5591 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
5592 (rs6000_insn_cost): New function.
5593 * config/rs6000/rs6000.md (cost): New attribute.
5595 2017-10-10 Jakub Jelinek <jakub@redhat.com>
5596 H.J. Lu <hongjiu.lu@intel.com>
5600 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
5601 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
5602 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
5603 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
5604 to OPTION_MASK_ISA_AVX512VL - builtins that have both
5605 OPTION_MASK_ISA_MMX and some other bit set require both
5606 mmx and the ISAs without the mmx bit.
5607 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
5608 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
5609 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
5610 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
5611 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
5612 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
5613 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
5614 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
5615 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
5616 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
5617 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
5618 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
5619 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
5620 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
5621 Add OPTION_MASK_ISA_MMX.
5623 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
5625 * config.gcc (armv7*-*-freebsd*): New target.
5626 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
5628 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
5630 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
5631 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
5634 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
5636 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
5637 a template parameter.
5638 (WIDE_INT_REF_FOR): Update accordingly.
5639 * tree.h (wi::int_traits <const_tree>): Delete.
5640 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
5641 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
5642 (wi::tree_to_wide_ref): New typedef.
5643 (wi::to_wide): New function.
5644 * calls.c (get_size_range): Use wi::to_wide when operating on
5646 * cgraph.c (cgraph_node::create_thunk): Likewise.
5647 * config/i386/i386.c (ix86_data_alignment): Likewise.
5648 (ix86_local_alignment): Likewise.
5649 * dbxout.c (stabstr_O): Likewise.
5650 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
5651 * expr.c (const_vector_from_tree): Likewise.
5652 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
5653 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
5654 (fold_negate_expr_1, int_const_binop_1, const_binop)
5655 (fold_convert_const_int_from_real, optimize_bit_field_compare)
5656 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
5657 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
5658 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
5659 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
5660 (fold_not_const, round_up_loc): Likewise.
5661 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
5662 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
5663 (alloca_call_type): Likewise.
5664 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
5665 * godump.c (go_output_typedef): Likewise.
5666 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
5667 * internal-fn.c (get_min_precision): Likewise.
5668 * ipa-cp.c (ipcp_store_vr_results): Likewise.
5669 * ipa-polymorphic-call.c
5670 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
5671 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
5672 (ipa_modify_call_arguments): Likewise.
5673 * match.pd: Likewise.
5674 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
5675 * print-tree.c (print_node_brief, print_node): Likewise.
5676 * stmt.c (expand_case): Likewise.
5677 * stor-layout.c (layout_type): Likewise.
5678 * tree-affine.c (tree_to_aff_combination): Likewise.
5679 * tree-cfg.c (group_case_labels_stmt): Likewise.
5680 * tree-data-ref.c (dr_analyze_indices): Likewise.
5681 (prune_runtime_alias_test_list): Likewise.
5682 * tree-dump.c (dequeue_and_dump): Likewise.
5683 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
5684 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
5685 * tree-pretty-print.c (dump_generic_node): Likewise.
5686 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
5687 (simple_iv_with_niters): Likewise.
5688 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
5689 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
5690 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
5691 * tree-ssa-loop-niter.c (split_to_var_and_offset)
5692 (refine_value_range_using_guard, number_of_iterations_ne_max)
5693 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
5694 (get_cst_init_from_scev, record_nonwrapping_iv)
5695 (scev_var_range_cant_overflow): Likewise.
5696 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
5697 * tree-ssa-pre.c (compute_avail): Likewise.
5698 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
5699 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
5700 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
5701 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
5702 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
5703 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
5704 (dump_case_nodes, try_switch_expansion): Likewise.
5705 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
5706 (vect_do_peeling): Likewise.
5707 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
5708 * tree-vect-stmts.c (vectorizable_load): Likewise.
5709 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
5710 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
5711 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
5712 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
5713 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
5714 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
5715 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
5716 (evrp_dom_walker::before_dom_children): Likewise.
5717 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
5718 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
5719 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
5720 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
5721 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
5722 (get_range_pos_neg): Likewise.
5723 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
5724 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
5725 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
5726 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
5727 * config/avr/avr.c (avr_fold_builtin): Likewise.
5728 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
5729 * config/msp430/msp430.c (msp430_attr): Likewise.
5730 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
5731 * config/powerpcspe/powerpcspe-c.c
5732 (altivec_resolve_overloaded_builtin): Likewise.
5733 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
5734 (rs6000_expand_ternop_builtin): Likewise.
5735 * config/rs6000/rs6000-c.c
5736 (altivec_resolve_overloaded_builtin): Likewise.
5737 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
5738 (rs6000_expand_ternop_builtin): Likewise.
5739 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5741 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5743 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5744 when copying loop nest with only one inner loop.
5746 2017-10-10 Richard Biener <rguenther@suse.de>
5748 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5749 blocks if SCEV is active.
5750 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5752 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5753 (scev_initialize): Assert we are not yet initialized.
5755 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5757 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5758 inner loop's exit stmt by making it always exit the loop, otherwise
5759 we would generate an infinite empty loop.
5761 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5763 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5764 renaming variables in new preheader if it's deleted.
5766 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5768 * tree-loop-distribution.c (struct partition): Remove unused field
5769 loops of the structure.
5770 (partition_alloc, partition_free): Ditto.
5771 (build_rdg_partition_for_vertex): Ditto.
5773 2017-10-09 Jeff Law <law@redhat.com>
5775 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
5776 return type to match prototype and documentation.
5778 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5780 * config/rs6000/rs6000.c (processor_costs): Move to ...
5781 * config/rs6000/rs6000.h: ... here.
5782 (rs6000_cost): Declare.
5784 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
5786 * except.c (setjmp_fn): New global variable.
5787 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5788 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
5789 if DONT_USE_BUILTIN_SETJMP is defined.
5791 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5793 * target.def (insn_cost): New hook.
5794 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
5795 * doc/tm.texi: Regenerate.
5796 * rtlanal.c (insn_cost): Use the new hook.
5798 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5800 * combine.c (combine_validate_cost): Compute the new insn_cost,
5801 not just pattern_cost.
5802 (try_combine): Adjust comment.
5804 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5806 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
5808 * combine.c (uid_insn_cost): Adjust comment.
5809 (combine_validate_cost): Adjust comment. Use pattern_cost instead
5811 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
5812 * dse.c (find_shift_sequence): Ditto.
5813 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
5814 (bb_valid_for_noce_process_p): Use pattern_cost.
5815 * rtl.h (insn_rtx_cost): Delete.
5816 (pattern_cost): New prototype.
5817 (insn_cost): New prototype.
5818 * rtlanal.c (insn_rtx_cost): Rename to...
5819 (pattern_cost): ... this.
5822 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5824 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5825 (*jcc<mode>_0_r_i387): Ditto.
5826 (*jccxf_r_i387): Ditto.
5827 (*jcc<mode>_r_i387): Ditto.
5828 (*jccu<mode>_r_i387): Ditto.
5829 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5830 (*jcc): Rename from *jcc_1.
5832 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5834 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5835 deferred rescans after the lvx/stvx recombination pre-pass.
5837 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5839 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5840 memory operation instruction support.
5841 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5842 (rs6000-ibm-aix[789]*): Likewise.
5843 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5844 Document new functions.
5846 2017-10-09 Richard Biener <rguenther@suse.de>
5848 PR tree-optimization/82397
5849 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5850 equality only for semantically equal trees.
5852 2017-10-09 Richard Biener <rguenther@suse.de>
5854 PR tree-optimization/82449
5855 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5856 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5857 allow constant addresses.
5858 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5861 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5863 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5866 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5869 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5872 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5875 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5877 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5880 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5881 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5883 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5885 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5886 (WI_BINARY_PREDICATE_RESULT): Likewise.
5887 (wi::binary_traits::operator_result): New type.
5888 (wi::binary_traits::predicate_result): Likewise.
5889 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5890 (generic_wide_int::operator==, generic_wide_int::operator!=)
5891 (generic_wide_int::operator&, generic_wide_int::and_not)
5892 (generic_wide_int::operator|, generic_wide_int::or_not)
5893 (generic_wide_int::operator^, generic_wide_int::operator+
5894 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5896 (operator~, unary operator-, operator==, operator!=, operator&)
5897 (operator|, operator^, operator+, binary operator-, operator*): New
5899 * expr.c (get_inner_reference): Use wi::bit_and_not.
5900 * fold-const.c (fold_binary_loc): Likewise.
5901 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5902 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5903 (bit_value_binop): Likewise.
5904 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5905 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5906 (extract_range_from_binary_expr_1): Likewise.
5907 (masked_increment): Likewise.
5908 (simplify_bit_ops_using_ranges): Likewise.
5910 2017-10-09 Martin Jambor <mjambor@suse.cz>
5913 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5914 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5915 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5917 (hsa_fixup_mov_insn_type): New function.
5918 (hsa_op_with_type::get_in_type): Use it.
5919 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5920 immediates in an assert.
5921 (hsa_op_with_type::extend_int_to_32bit): New method.
5922 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5923 types. Convert to dest type if necessary.
5924 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5925 (reg_for_gimple_ssa): Pass false as min32int to
5926 hsa_type_for_scalar_tree_type.
5927 (gen_hsa_addr): Fixup type when creating addresable temporary.
5928 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5929 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5930 necessary. Call hsa_fixup_mov_insn_type.
5931 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5932 extend operands and convert to dest type if necessary.
5933 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5934 to dest type if necessary.
5935 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5936 if conversion nt necessary and size matches.
5937 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5938 to dest type if necessary.
5939 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5940 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5942 (gen_hsa_clrsb): Likewise.
5943 (gen_hsa_ffs): Likewise.
5944 (gen_hsa_divmod): Extend operands and convert to dest type if
5946 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5948 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5950 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5951 Remove empty default arguments. Use a brace block as output
5953 (conditional return): Ditto.
5955 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5956 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5958 (group_ending_nop): Ditto.
5959 (doloop_end): Ditto.
5960 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5961 (splitters for those): Ditto.
5963 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5965 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5966 a conditional jump (and the compare for it) so that pc_rtx is the
5968 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5969 for the deleted and renamed ctr<mode>_internal[234] patterns.
5970 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5971 Delete second conditional return pattern.
5972 (ctr<mode>_internal2): Delete this second bdnz pattern.
5973 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5974 (ctr<mode>_internal4): Delete this second bdz pattern.
5976 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5978 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5979 (always_initialized_rtx_for_ssa_name_p): New predicate.
5980 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5981 (finish_out_of_ssa): Free new field of SA.
5982 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5983 * tree-ssa-coalesce.c: Include tree-ssa.h.
5984 (get_parm_default_def_partitions): Remove extern keyword.
5985 (get_undefined_value_partitions): New function.
5986 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5987 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5989 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5991 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5993 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5995 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5997 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5998 for targets that preffer 128bit.
6000 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6002 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6004 2017-10-08 Olivier Hainque <hainque@adacore.com>
6006 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6007 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6009 (thumb_set_return_address): Likewise.
6011 2017-10-08 Olivier Hainque <hainque@adacore.com>
6013 * common/config/arm/arm-common.c (arm_except_unwind_info):
6014 Handle DWARF2_UNWIND_INFO.
6016 2017-10-07 Michael Collison <michael.collison@arm.com>
6018 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6021 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6023 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6024 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6025 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6026 defined, force the creation of a new block for a dispatch label.
6028 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6030 * invoke.texi (Wsuggest-attribute=cold): Document.
6031 * common.opt (Wsuggest-attribute=cold): New
6032 * ipa-pure-const.c (warn_function_cold): New function.
6033 * predict.c (compute_function_frequency): Use it.
6034 * predict.h (warn_function_cold): Declare.
6036 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6038 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6041 2017-10-06 Martin Liska <mliska@suse.cz>
6043 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6044 keyword for member functions.
6045 (struct sanopt_tree_couple): New struct.
6046 (struct sanopt_tree_couple_hash): New function.
6047 (struct sanopt_ctx): Add new hash_map.
6048 (has_dominating_ubsan_ptr_check): New function.
6049 (record_ubsan_ptr_check_stmt): Likewise.
6050 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6051 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6052 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6054 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6057 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6058 aarch64_simd_valid_immediate on CONST_VECTORs.
6059 (aarch64_reg_or_bic_imm): Likewise.
6061 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6063 PR rtl-optimization/82396
6064 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6066 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6068 * graphite-dependences.c (scop_get_reads): Move code to...
6069 (scop_get_must_writes): Move code to...
6070 (scop_get_may_writes): Move code to...
6071 (scop_get_reads_and_writes): ... here.
6072 (scop_get_dependences): Call scop_get_reads_and_writes.
6074 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6076 PR tree-optimization/82434
6077 * fold-const.h (can_native_encode_type_p,
6078 can_native_encode_string_p): Remove.
6079 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6080 don't encode anything, just return what would be otherwise returned.
6081 (native_encode_fixed, native_encode_complex, native_encode_vector):
6083 (native_encode_string): Likewise. Inline by hand
6084 can_native_encode_string_p.
6085 (can_native_encode_type_p): Remove.
6086 (can_native_encode_string_p): Remove.
6087 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6088 STRING_CSTs using can_native_encode_string_p, test all
6089 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6090 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6091 argument from native_encode_expr.
6092 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6093 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6096 2017-10-06 Richard Biener <rguenther@suse.de>
6098 PR tree-optimization/82397
6099 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6100 operand_equal_p but rely on data_ref_compare_tree for detecting
6102 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6103 to match up with dr_group_sort_cmp.
6105 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6108 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6110 * config/s390/s390-builtin-types.def: Regenerate.
6112 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6115 * config/s390/s390-builtin-types.def: Regenerate.
6116 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6117 Change flag from B_VXE to B_VX.
6118 (s390_vec_min_dbl): Remove B_VXE flag.
6120 2017-10-06 Richard Biener <rguenther@suse.de>
6122 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6123 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6124 translate_isl_ast_to_gimple::is_valid_rename,
6125 translate_isl_ast_to_gimple::get_rename,
6126 translate_isl_ast_to_gimple::get_def_bb_for_const,
6127 translate_isl_ast_to_gimple::get_new_name,
6128 translate_isl_ast_to_gimple::collect_all_ssa_names,
6129 translate_isl_ast_to_gimple::copy_loop_phi_args,
6130 translate_isl_ast_to_gimple::collect_all_ssa_names,
6131 translate_isl_ast_to_gimple::copy_loop_phi_args,
6132 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6133 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6134 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6135 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6136 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6137 translate_isl_ast_to_gimple::copy_cond_phi_args,
6138 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6139 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6140 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6141 translate_isl_ast_to_gimple::rename_uses,
6142 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6143 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6144 (set_rename_for_each_def): Likewise.
6145 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6146 here. Handle rewriting SCEV analyzable uses here.
6147 (copy_bb_and_scalar_dependences): Generate code for PHI
6149 (graphite_regenerate_ast_isl): Adjust.
6150 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6151 (add_write, add_read): New functions.
6152 (build_cross_bb_scalars_def): Use it and simplify.
6153 (build_cross_bb_scalars_use): Likewise.
6154 (graphite_find_cross_bb_scalar_vars): Inline into...
6155 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6156 simulating out-of-SSA. Compute liveout and add dependencies.
6157 (build_scops): Force an empty entry block.
6158 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6160 (sese_build_liveouts): Declare.
6161 (sese_trivially_empty_bb_p): Likewise.
6162 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6163 compute liveout and debug_liveout.
6164 (sese_bad_liveouts_use): Remove.
6165 (sese_reset_debug_liveouts_bb): Likewise.
6166 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6167 (sese_build_liveouts): Build liveout and debug_liveout and store
6169 (new_sese_info): Adjust.
6170 (free_sese_info): Likewise.
6171 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6172 do not build liveout here.
6173 (move_sese_in_condition): Adjust region entry.
6174 (scev_analyzable_p): Match up with chrec_apply requirements.
6175 (sese_trivially_empty_bb_p): New.
6176 * tree-into-ssa.c (get_reaching_def): Properly support generating
6177 default-defs for incremental rewrite of anonymous names.
6179 2017-10-06 Richard Biener <rguenther@suse.de>
6181 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6182 precision do not perform modulo reduction.
6184 2017-10-06 Richard Biener <rguenther@suse.de>
6186 PR tree-optimization/82436
6187 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6188 conservatively choose the vectorization factor when checking
6189 whether we can perform the required load permutation.
6190 (vect_transform_slp_perm_load): Assert when we may not fail.
6192 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6194 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6195 message for incompatible -msdata=* and -mcall-* options.
6197 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6199 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6200 rate for post-reload scheduling.
6202 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6204 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6206 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6208 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6209 to improve monte carlo in scimark.
6211 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6213 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6214 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6215 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6216 pentium4_cost, nocona_cost): Set reassociation width to 1.
6217 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6218 width to 2 for fp operations and 1 otherwise.
6219 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6221 (atom_cost): Set reassociation width to 2.
6222 (slm_cost, generic_cost): Set fp reassociation width
6223 to 2 and 1 otherwise.
6224 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6225 (core_cost): Set fp reassociation width to 4 and vector to 2.
6226 (ix86_reassociation_width): Rewrite using cost table; special case
6227 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6228 and TARGET_AVX128_OPTIMAL.
6229 * config/i386/i386.h (processor_costs): Add
6230 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6231 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6232 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6233 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6234 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6235 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6237 2017-10-05 Nathan Sidwell <nathan@acm.org>
6239 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6241 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6243 * config/arm/arm.c (arm_test_fpu_data): New.
6244 (arm_run_selftests): Call arm_test_fpu_data.
6246 2017-10-04 Nathan Sidwell <nathan@acm.org>
6248 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6250 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6252 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6254 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6255 check type for aarch64_simd_valid_immediate.
6256 (aarch64_output_simd_mov_immediate): Update prototype.
6257 (aarch64_simd_valid_immediate): Update prototype.
6258 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6259 support for ORR-immediate.
6260 (and<mode>3): modified pattern to add support for BIC-immediate.
6261 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6262 now checks for valid immediate for BIC and ORR based on new enum
6264 (aarch64_output_simd_mov_immediate): Function now used to output
6265 BIC/ORR imm as well based on new enum argument.
6266 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6268 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6269 (aarch64_reg_or_bic_imm): Likewise.
6271 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6273 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6274 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6276 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6280 PR rtl-optimization/82396
6281 * haifa-sched.c (autopref_multipass_init): Simplify
6283 (autopref_rank_data): Simplify sort order.
6284 * sched-int.h (autopref_multipass_data_): Remove
6285 multi_mem_insn_p, min_offset and max_offset.
6287 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6289 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6291 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6293 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6296 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6298 * doc/sourcebuild.texi: Document vect_long_mult.
6300 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6302 PR tree-optimization/82413
6303 * fold-const.c (build_range_check): Use widest_int when comparing
6304 the maximum ETYPE value with HIGH.
6306 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6308 PR rtl-optimization/82396
6309 * haifa-sched.c (autopref_multipass_init): Simplify
6311 (autopref_rank_data): Simplify sort order.
6312 * sched-int.h (autopref_multipass_data_): Remove
6313 multi_mem_insn_p, min_offset and max_offset.
6315 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6317 PR tree-optimization/82381
6318 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6319 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6322 PR tree-optimization/82374
6323 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6324 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6325 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6326 current_function_decl to the new decl.
6328 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6330 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6331 helper macro for IEEE float128 hardware built-in functions.
6332 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6334 (TRUNCF128_ODD): Likewise.
6335 (ADDF128_ODD): Likewise.
6336 (SUBF128_ODD): Likewise.
6337 (MULF128_ODD): Likewise.
6338 (DIVF128_ODD): Likewise.
6339 (FMAF128_ODD): Likewise.
6340 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6341 UNSPEC_TRUNC_ROUND_TO_ODD.
6342 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6343 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6344 floating point round to odd instructions.
6345 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6346 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6347 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6348 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6349 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6350 (trunc<mode>sf2_hw): Change the truncate with round to odd
6351 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6352 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6353 to odd hardware instructions.
6354 (sub<mode>3_odd): Likewise.
6355 (mul<mode>3_odd): Likewise.
6356 (div<mode>3_odd): Likewise.
6357 (sqrt<mode>2_odd): Likewise.
6358 (fma<mode>4_odd): Likewise.
6359 (fms<mode>4_odd): Likewise.
6360 (nfma<mode>4_odd): Likewise.
6361 (nfms<mode>4_odd): Likewise.
6362 (trunc<mode>df2_odd): Change the truncate with round to odd
6363 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6365 * doc/extend.texi (PowerPC built-in functions): Update documentation
6366 for existing IEEE float128-bit built-in functions. Add built-in
6367 functions that generate the IEEE 128-bit floating point round to
6370 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6372 PR rtl-optimization/77729
6373 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6374 to (X&(C1&~C2))|C2 transformations.
6376 2017-10-03 Martin Jambor <mjambor@suse.cz>
6378 PR tree-optimization/82363
6379 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6380 mismatch, mark lacc written regardless of racc.
6382 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6384 PR tree-optimization/82381
6385 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6386 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6387 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6388 fallthrough into reassoc_stmt_dominates_stmt_p.
6391 * combine.c (combine_instructions): Don't combine in unreachable
6394 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6397 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6398 function to not use the have_cpu variable. Do not set cpu_index,
6399 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6401 (rs6000_valid_attribute_p): Remove duplicate initializations of
6402 old_optimize and func_optimize.
6403 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6404 (rs6000_activate_target_options): Make global.
6405 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6408 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6410 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6411 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6412 Set *poffset to 0 and *psize and *pmax_size to -1 if
6413 *poffset + *psize overflows in HOST_WIDE_INT.
6415 PR tree-optimization/82387
6416 PR tree-optimization/82388
6417 PR tree-optimization/82389
6418 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6419 instead of live_bytes non-NULL.
6421 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6424 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6427 2017-10-02 Richard Biener <rguenther@suse.de>
6429 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6430 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6431 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6433 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6435 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6436 requested precision matches the type's.
6437 * calls.c (alloc_max_size): Calculate the new candidate size as
6438 a widest_int and use wi::to_widest when comparing it with the
6439 current candidate size.
6440 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6441 zero rather than integer_zero_node.
6442 * match.pd: Check for a no-op conversion before using wi::add
6443 rather than after. Use tree_to_uhwi when summing small shift
6444 counts into an unsigned int.
6446 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6447 Alan Hayward <alan.hayward@arm.com>
6448 David Sherwood <david.sherwood@arm.com>
6451 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6452 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6453 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6454 POINTER_AND_FP_REGS.
6456 2017-10-02 Richard Biener <rguenther@suse.de>
6458 PR tree-optimization/82355
6459 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6460 a mapping for the enclosing loop but avoid generating one for
6462 (copy_bb_and_scalar_dependences): Remove premature codegen
6463 error on PHIs in blocks duplicated into multiple places.
6464 * graphite-scop-detection.c
6465 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6466 in the region use it as loop and nest to analyze the DR in.
6467 (try_generate_gimple_bb): Likewise.
6468 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6469 (add_loop_constraints): For blocks in a loop not in the region
6470 create a dimension with a single iteration.
6471 * sese.h (gbb_loop_at_index): Remove assert.
6473 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6475 * omp-expand.c (adjust_context_scope): New function.
6476 (expand_parallel_call): Call adjust_context_scope.
6478 2017-10-01 Jeff Law <law@redhat.com>
6480 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6481 dom_opt_dom_walker class with direct access to private members.
6482 Add comments. Call test_for_singularity.
6483 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6484 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6485 m_dummy_cond anymore.
6486 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6488 (pass_dominator:execute): Build the dummy_cond here and pass it
6489 to the dom_opt_dom_walker ctor.
6490 (test_for_singularity): New function.
6492 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6493 Maya Rashish <coypu@sdf.org>
6495 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6496 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6497 (alpha*-*-netbsd*) Use nbsd_tm_file.
6498 (arm*-*-netbsdelf*) Likewise.
6499 (i[34567]86-*-netbsdelf*) Likewise.
6500 (x86_64-*-netbsd*) Likewise.
6501 (mips*-*-netbsd*) Likewise.
6502 (powerpc-*-netbsd*) Likewise.
6503 (sh*-*-netbsd*) Likewise.
6504 (sparc-*-netbsdelf*) Likewise.
6505 (sparc64-*-netbsd*) Likewise.
6506 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6508 (vax-*-netbsdelf*) Likewise.
6509 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6510 (UINT_FAST8_TYPE) Likewise.
6511 (INT_FAST16_TYPE) Check CHAR_FAST16.
6512 (UINT_FAST16_TYPE) Likewise.
6514 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6517 * config/i386/i386.md
6518 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6519 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6520 *divmodsi4_zext_2): New define_insn_and_split.
6521 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6522 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6523 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6524 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6525 New define_insn_and_split.
6526 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6527 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6528 operands[1] having DImode when mode is SImode.
6530 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6531 always SImode for DIV and MOD in REG_EQUAL notes.
6533 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6536 * match.pd: Fix handling of NaNs in pattern.
6538 2017-09-29 Jeff Law <law@redhat.com>
6540 * sbitmap.c (bitmap_bit_in_range_p): New function.
6541 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6542 * tree-ssa-dse.c (live_bytes_read): New function.
6543 (dse_classify_store): Ignore reads of dead bytes.
6545 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6546 typos and whitespace errors.
6547 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6548 * config/s390/s390.c (s390_emit_prologue): Likewise.
6550 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6553 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6554 with a symbol for LRA.
6556 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6558 PR rtl-optimization/82338
6559 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
6561 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
6563 * genmodes.c (calc_wider_mode): Suppress qsort macro.
6564 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
6565 (qsort_chk): Declare.
6566 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
6567 (qsort_chk): New function.
6569 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6571 PR tree-optimization/82337
6572 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
6573 phi definition if the PHI result appears in an abnormal PHI.
6574 (find_basis_for_base_expr): Don't record a basis if the LHS of the
6575 basis appears in an abnormal PHI.
6577 2017-09-29 Richard Biener <rguenther@suse.de>
6579 * graphite-isl-ast-to-gimple.c
6580 (translate_isl_ast_to_gimple::set_codegen_error): New function.
6581 (binary_op_to_tree): Use it.
6582 (get_rename_from_scev): Likewise.
6583 (copy_loop_phi_nodes): Likewise.
6584 (copy_bb_and_scalar_dependences): Likewise.
6585 (translate_pending_phi_nodes): Likewise.
6587 2017-09-29 Jakub Jelinek <jakub@redhat.com>
6590 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
6591 for movabsq $(i32 << shift), r64.
6593 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
6595 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
6596 index when encoding %esp as %rsp to avoid 0x67 prefix.
6598 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
6600 * config/i386/i386.md (*movsf_internal, *movdf_internal):
6601 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
6603 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
6605 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
6606 Extensions with more than 16 double VFP registers.
6607 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6608 to_clear_mask and all code related to it. Replace the remaining
6609 entry by a sbitmap and adapt code accordingly.
6611 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
6613 * brig-builtins.def: Change pure attributes to const.
6615 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6617 * config.gcc (default_gnu_indirect_function): Default to yes for
6618 sparc*-*-linux* with glibc.
6620 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6622 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
6623 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
6624 when creating .init_array and .fini_array sections with priority
6627 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
6630 * config/aarch64/aarch64.c
6631 (aarch64_builtin_support_vector_misalignment): Always return false
6632 when misalignment is unknown.
6634 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
6636 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
6637 this function to return false if the definition used by the swap
6638 instruction is artificial, or if the memory address from which the
6639 constant value is loaded is not represented by a base address held
6640 in a register or if the base address register is a frame or stack
6641 pointer. Additionally, return false if the base address of the
6642 loaded constant is a SYMBOL_REF but is not considered to be a
6644 (replace_swapped_load_constant): New function.
6645 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
6646 loaded constant vector with a load of a swapped constant vector.
6648 2017-09-27 Carl Love <cel@us.ibm.com>
6650 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
6651 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
6652 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
6655 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
6657 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
6658 first, always call autopref_rank_data otherwise.
6660 2017-09-27 Richard Biener <rguenther@suse.de>
6662 * graphite-scop-detection.c (find_scop_parameters): Move
6663 loop bound handling ...
6664 (gather_bbs::before_dom_children): ... here, avoiding the need
6665 to build scop_info->loop_nest.
6666 (record_loop_in_sese): Remove.
6667 * sese.h (sese_info_t::loop_nest): Remove.
6668 * sese.c (new_sese_info): Do not allocate loop_nest.
6669 (free_sese_info): Do not free loop_nest.
6671 2017-09-27 Jakub Jelinek <jakub@redhat.com>
6674 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
6675 lhs from calls if the lhs has addressable type.
6677 2017-09-27 Richard Biener <rguenther@suse.de>
6679 * graphite.h (scop::max_alias_set): New member.
6680 * graphite-scop-detection.c: Remove references to non-existing
6681 --param in comments.
6682 (build_alias_sets): Record the maximum alias set used for drs.
6683 (build_scops): Support zero as unlimited for
6684 --param graphite-max-arrays-per-scop.
6685 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
6687 (build_poly_sr_1): ... here. Compute alias set based on the
6688 maximum alias set used for drs rather than
6689 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
6691 2017-09-27 Richard Biener <rguenther@suse.de>
6693 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
6694 --param loop-block-tile-size=0 to disable tiling.
6696 2017-09-27 Richard Biener <rguenther@suse.de>
6698 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
6699 (graphite-max-nb-scop-params): Document special value zero.
6700 * domwalk.h (dom_walker::STOP): New symbolical constant.
6701 (dom_walker::dom_walker): Add optional parameter for bb to
6703 (dom_walker::~dom_walker): Declare.
6704 (dom_walker::before_dom_children): Document STOP return value.
6705 (dom_walker::m_user_bb_to_rpo): New member.
6706 (dom_walker::m_bb_to_rpo): Likewise.
6707 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
6708 mapping here if not provided by the user.
6709 (dom_walker::~dom_walker): Free bb to RPO mapping if not
6710 provided by the user.
6711 (dom_walker::STOP): Define.
6712 (dom_walker::walk): Do not compute bb to RPO mapping here.
6713 Support STOP return value from before_dom_children to stop
6715 * graphite-optimize-isl.c (optimize_isl): If the schedule
6716 is the same still generate code if -fgraphite-identity
6717 or -floop-parallelize-all are given.
6718 * graphite-scop-detection.c: Include cfganal.h.
6719 (gather_bbs::gather_bbs): Get and pass through bb to RPO
6721 (gather_bbs::before_dom_children): Return STOP for BBs
6723 (build_scops): Compute bb to RPO mapping and pass it to
6724 the domwalk. Treat --param graphite-max-nb-scop-params=0
6725 as not limiting the number of params.
6726 * graphite.c (graphite_initialize): Remove limit on the
6727 number of basic-blocks in a function.
6728 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
6729 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
6730 default value of 10.
6732 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6734 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
6735 Adjust code to eliminate needing to do the shift right 32-bits
6736 operation after XSCVDPSPN.
6738 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6740 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6741 ((X / Y) != 0 -> X >= Y): Likewise.
6743 2017-09-26 Carl Love <cel@us.ibm.com>
6745 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6746 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6747 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6748 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6749 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6750 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6751 definitions and overloading.
6752 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6753 statement for P9V_BUILTIN_XST_LEN_R.
6754 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6755 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6756 define_expand and define_insn for the instructions and builtins.
6757 * doc/extend.texi: Update the built-in documentation file for the new
6759 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6760 define_insn for the instructions
6762 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6765 * gcc/config/netbsd-protos.h: New file.
6766 * gcc/config/netbsd.c: New file.
6767 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
6768 * gcc/config/t-netbsd: New file.
6769 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
6770 (tmake_file) Add t-netbsd.
6771 (extra_objs) Add netbsd.o.
6773 2017-09-26 Janus Weil <janus@gcc.gnu.org>
6777 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
6779 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6781 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
6782 sign extension from a vector register to a GPR by doing a 32-bit
6783 direct move and then an EXTSW.
6784 (extendsi<mode>2 splitter): Likewise.
6785 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
6786 right or vector extract after doing XSCVDPSPN. Use
6787 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
6789 (movdi_from_sf_zero_ext): Likewise.
6790 (reload_gpr_from_vsxsf): Likewise.
6791 (p8_mfvsrd_4_disf): Delete, no longer used.
6792 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
6793 then needing to move the SFmode to a GPR to use the XSCVDPSP
6794 instruction instead of FRSP and XSCVDPSPN.
6795 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
6796 it is adjacent to the other XSCVSPDP insns.
6797 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
6798 SFmode to be in traditional Altivec registers.
6799 (vsx_xscvdpspn): Eliminate useless alternative constraint.
6800 (vsx_xscvspdpn): Likewise.
6801 (vsx_xscvspdpn_scalar): Likewise.
6803 2017-09-26 Martin Jambor <mjambor@suse.cz>
6805 * tree-sra.c (compare_access_positions): Put integral types first,
6806 stabilize sorting of integral types, remove conditions putting
6807 non-full-precision integers last.
6808 (sort_and_splice_var_accesses): Disable scalarization if a
6809 non-integert would be represented by a non-full-precision integer.
6811 2017-09-26 Joseph Myers <joseph@codesourcery.com>
6813 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
6814 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
6815 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6816 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
6817 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
6818 conditionals inside the function instead of around it. Call
6819 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6820 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6822 2017-09-26 Richard Biener <rguenther@suse.de>
6824 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6826 (scop_detection::build_scop_breadth): ... this. Removed.
6827 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6828 (scop_detection::harmful_stmt_in_bb): Likewise.
6829 (scop_detection::graphite_can_represent_stmt): Likewise.
6830 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6831 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6832 (scop_detection::can_represent_loop_1): ... this. Removed.
6833 (scop_detection::harmful_loop_in_region): Simplify after inlining
6834 the above and remove more quadraticness.
6835 (build_scops): Adjust.
6836 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6839 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6842 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6843 REGNO (base) == SP_REG if base is a REG.
6846 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6847 if it is different SSA_NAME.
6848 (optimize_range_tests_cmp_bitwise): New function.
6849 (optimize_range_tests): Call it.
6851 2017-09-26 Richard Biener <rguenther@suse.de>
6853 PR tree-optimization/82321
6854 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6855 for the def being inside the loop.
6857 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6859 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6861 * config/s390/s390-builtins.def: Fix constraint on op4.
6863 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6865 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6866 independent expanders.
6867 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6868 ("vec_ordered", "vec_unordered"): New expanders.
6870 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6872 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6875 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6877 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6878 vec_unpacks_lo_v16qi.
6879 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6881 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6883 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6884 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6885 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6887 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6889 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6891 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6893 ("*vec_slb<mode>"): New insn pattern.
6894 ("vec_shr_<mode>"): New expander.
6895 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6896 and force the shift count operand to V16QImode.
6897 ("vec_srb<mode>"): Set shift count mode to V16QI.
6899 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6901 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6902 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6903 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6905 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6908 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6910 2017-09-26 Richard Biener <rguenther@suse.de>
6912 PR tree-optimization/82320
6913 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6916 2017-09-25 Jeff Law <law@redhat.com>
6918 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6919 prototype for new argument.
6920 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6921 mostly extracted from rs6000_emit_allocate_stack.
6922 (rs6000_emit_probe_stack_range_stack_clash): New function.
6923 (rs6000_emit_allocate_stack): Call
6924 rs6000_emit_probe_stack_range_stack_clash as needed.
6925 (rs6000_emit_probe_stack_range): Add additional argument
6926 to call to gen_probe_stack_range{si,di}.
6927 (output_probe_stack_range): New.
6928 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6929 (output_probe_stack_range_stack_clash): New.
6930 (rs6000_emit_prologue): Emit notes into dump file as requested.
6931 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6932 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6933 Add additional operand and pass it to output_probe_stack_range.
6935 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6937 PR tree-optimization/82163
6938 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6939 (checking_verify_loop_closed_ssa): New parameter.
6940 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6941 (check_loop_closed_ssa_stmt): Delete.
6942 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6943 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6944 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6947 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6949 * brig-builtins.def: Treat HSAIL barrier builtins as
6950 setjmp/longjump style functions.
6952 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6954 * target.def (constant_alignment): New hook.
6955 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6956 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6957 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6958 * doc/tm.texi: Regenerate.
6959 * targhooks.h (default_constant_alignment): Declare.
6960 (constant_alignment_word_strings): Likewise.
6961 * targhooks.c (default_constant_alignment): New function.
6962 (constant_alignment_word_strings): Likewise.
6963 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6964 instead of CONSTANT_ALIGNMENT.
6965 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6966 (force_const_mem): Likewise.
6967 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6968 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6969 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6970 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6971 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6973 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6974 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6975 constant_alignment_word_strings.
6976 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6977 (CONSTANT_ALIGNMENT): Likewise.
6978 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6979 (arm_constant_alignment): New function.
6980 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6981 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6982 constant_alignment_word_strings.
6983 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6984 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6985 constant_alignment_word_strings.
6986 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6987 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6988 (cris_constant_alignment): New function.
6989 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6990 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6991 (epiphany_constant_alignment): New function.
6992 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6993 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6994 constant_alignment_word_strings.
6995 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6996 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6997 constant_alignment_word_strings.
6998 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6999 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7000 constant_alignment_word_strings.
7001 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7002 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7003 * config/i386/i386.c (ix86_constant_alignment): Make static.
7004 Use the same interface as the target hook.
7005 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7006 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7007 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7008 constant_alignment_word_strings.
7009 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7010 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7011 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7012 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7013 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7014 constant_alignment_word_strings.
7015 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7016 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7017 constant_alignment_word_strings.
7018 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7019 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7020 constant_alignment_word_strings.
7021 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7022 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7024 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7025 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7026 * config/mips/mips.c (mips_constant_alignment): New function.
7027 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7028 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7029 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7030 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7031 (mmix_constant_alignment): Make static. Use the same interface
7033 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7034 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7035 constant_alignment_word_strings.
7036 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7037 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7038 constant_alignment_word_strings.
7039 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7040 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7041 constant_alignment_word_strings.
7042 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7043 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7044 (rs6000_constant_alignment): New function.
7045 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7046 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7047 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7048 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7049 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7050 (rs6000_constant_alignment): New function.
7051 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7052 * config/s390/s390.c (s390_constant_alignment): New function.
7053 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7054 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7055 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7056 constant_alignment_word_strings.
7057 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7058 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7059 (sparc_constant_alignment): New function.
7060 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7061 * config/spu/spu.c (spu_constant_alignment): New function.
7062 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7063 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7064 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7065 constant_alignment_word_strings.
7066 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7067 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7068 constant_alignment_word_strings.
7069 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7070 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7071 constant_alignment_word_strings.
7072 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7073 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7074 (visium_constant_alignment): New function.
7075 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7076 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7077 (xtensa_constant_alignment): New function.
7078 * system.h (CONSTANT_ALIGNMENT): Poison.
7080 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7082 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7083 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7084 (rs6000_builtin_valid_without_lhs): New helper function.
7085 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7086 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7088 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7090 * target.h (vec_perm_indices): Use unsigned short rather than
7092 (auto_vec_perm_indices): Likewise.
7093 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7094 Use unsigned int rather than unsigned char.
7095 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7097 2017-09-25 Richard Biener <rguenther@suse.de>
7099 * cfgloop.h (sort_sibling_loops): Declare.
7100 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7101 (sort_sibling_loops): New function sorting the sibling loop list
7103 * graphite.c (graphite_transform_loops): Sort sibling loops.
7105 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7107 * target.def (vec_perm_const_ok): Change sel parameter to
7109 * optabs-query.c (can_vec_perm_p): Update accordingly.
7110 * doc/tm.texi: Regenerate.
7111 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7112 auto_vec_perm_indices and remove separate nelt field.
7113 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7114 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7115 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7116 (aarch64_expand_vec_perm_const): Update accordingly.
7117 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7118 to vec_perm_indices.
7119 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7120 auto_vec_perm_indices and remove separate nelt field.
7121 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7122 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7123 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7125 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7126 to vec_perm_indices.
7127 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7128 sel to vec_perm_indices.
7129 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7130 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7131 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7133 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7136 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7139 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7140 on the FUNCTION_DECL function context if it has a DIE that is a
7143 2017-09-25 Richard Biener <rguenther@suse.de>
7145 PR tree-optimization/82285
7146 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7149 2017-09-25 Tom de Vries <tom@codesourcery.com>
7153 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7156 2017-09-25 Richard Biener <rguenther@suse.de>
7158 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7159 ISL errors other than isl_error_quota happen. Dump if the
7160 schedule is the same.
7161 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7162 errors instead of aborting inside ISL.
7164 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7167 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7168 of libgcc_eh for m64.
7169 * config/i386/darwin64.h: Likewise.
7171 2017-09-25 Richard Biener <rguenther@suse.de>
7174 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7175 attribute for incomplete types nor twice for complete ones.
7177 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7180 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7181 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7184 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7187 * config/i386/i386.opt (mprefer-avx256): Use
7188 ix86_target_flags variable.
7189 * config/i386/i386.c (ix86_target_string): Move
7190 -mprefer-avx256 to flag2_opts.
7192 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7195 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7196 and x != -1 | y != -1 into (x & y) != -1.
7198 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7200 * config.gcc: Add new case statement to set
7201 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7202 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7203 s390x-*-linux* case statements. Added aarch64 to the list of
7204 supported architectures.
7206 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7208 PR tree-optimization/82289
7209 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7210 STMT_VINFO_RELEVANT_P.
7212 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7213 Alan Hayward <alan.hayward@arm.com>
7214 David Sherwood <david.sherwood@arm.com>
7216 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7217 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7218 (extract_range_from_binary_expr_1): Don't call
7219 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7221 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7222 Alan Hayward <alan.hayward@arm.com>
7223 David Sherwood <david.sherwood@arm.com>
7225 * target.def (preferred_vector_alignment): New hook.
7226 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7228 * doc/tm.texi: Regenerate.
7229 * targhooks.h (default_preferred_vector_alignment): Declare.
7230 * targhooks.c (default_preferred_vector_alignment): New function.
7231 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7233 (DR_TARGET_ALIGNMENT): New macro.
7234 (aligned_access_p): Update commentary.
7235 (vect_known_alignment_in_bytes): New function.
7236 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7238 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7239 Calculate the misalignment based on the target alignment rather than
7241 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7242 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7243 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7244 the target alignment, rather than masking the element misalignment
7245 with the number of elements in a vector. Also use the target
7246 alignment when calculating the maximum number of peels.
7247 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7248 instead of TYPE_ALIGN_UNIT.
7249 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7250 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7251 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7252 (vect_create_data_ref_ptr): Likewise.
7253 (vect_setup_realignment): Realign by ANDing with
7254 -DR_TARGET_MISALIGNMENT.
7255 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7256 the number of peels based on DR_TARGET_ALIGNMENT.
7257 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7258 with the guaranteed alignment boundary when deciding whether
7260 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7261 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7262 (ensure_base_align): Remove stmt_info parameter. Get the
7263 target base alignment from DR_TARGET_ALIGNMENT.
7264 (vectorizable_store): Update call accordingly. Interpret
7265 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7267 (vectorizable_load): Likewise.
7269 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7270 Alan Hayward <alan.hayward@arm.com>
7271 David Sherwood <david.sherwood@arm.com>
7273 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7274 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7275 (vect_enhance_data_refs_alignment): Likewise.
7277 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7279 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7280 error. Only quit immediately if parsing is complete.
7281 (BEGIN): Initialize fatal_err and parse_done.
7282 (begin fpu, end fpu): Check number of arguments.
7283 (begin arch, end arch): Likewise.
7284 (begin cpu, end cpu): Likewise.
7285 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7286 (optalias): Likewise.
7288 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7290 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7291 * config/arm/arm-isa.h: Delete. Move definitions to ...
7292 * arm-cpus.in: ... here. Use new feature and fgroup values.
7293 * config/arm/arm.c (arm_option_override): Use lower case for feature
7295 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7296 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7297 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7299 (print_isa_bits_for): New function.
7300 (gen_isa): New function.
7301 (gen_comm_data): Use print_isa_bits_for.
7302 (define feature): New keyword.
7303 (define fgroup): New keyword.
7304 * config/arm/t-arm (TM_H): Remove.
7305 (GTM_H): Add arm-isa.h.
7306 (arm-isa.h): Add rule to generate file.
7307 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7308 case for feature bit names.
7310 2017-09-22 Richard Biener <rguenther@suse.de>
7312 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7314 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7315 print of no dependency loops ...
7316 * graphite.c (graphite_transform_loops): ... here.
7317 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7319 (same_close_phi_node, remove_duplicate_close_phi,
7320 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7321 (canonicalize_loop_closed_ssa): ... here and simplify.
7322 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7323 (optimize_isl): Use dump_printf_loc to tell when we stopped
7324 optimizing because of an ISL timeout.
7326 2017-09-22 Richard Biener <rguenther@suse.de>
7328 PR tree-optimization/82291
7329 * tree-if-conv.c (predicate_mem_writes): Make sure to
7330 remove writes in blocks predicated with false.
7332 2017-09-22 Richard Biener <rguenther@suse.de>
7334 * sese.c: Include cfganal.h.
7335 (if_region_set_false_region): Remove.
7336 (create_if_region_on_edge): Likewise.
7337 (move_sese_in_condition): Re-implement without destroying
7340 2017-09-22 Richard Biener <rguenther@suse.de>
7342 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7343 Verify both BBs contain loop PHI nodes before dispatching to
7345 (graphite_regenerate_ast_isl): Do not recompute dominators,
7346 do not verify three times. Restructure for clarity.
7347 * graphite-scop-detection.c (same_close_phi_node,
7348 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7349 defined_in_loop_p, canonicalize_loop_closed_ssa,
7350 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7351 checking and SSA rewrite, move to ...
7352 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7353 (graphite_initialize): Do not pass in ctx, do not reset the
7354 SCEV cache, compute only dominators.
7355 (graphite_transform_loops): Allocate ISL ctx after
7356 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7357 Maintain post-dominators only around build_scops.
7358 * sese.c (if_region_set_false_region): Make static. Free
7359 and recompute dominators.
7360 (move_sese_in_condition): Assert we don't get called with
7361 post-dominators computed.
7362 * sese.h (if_region_set_false_region): Remove.
7364 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7366 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7367 mode attribute for TARGET_AVX512VL.
7369 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7371 * config/i386/i386.opt (mprefer-avx256): New option.
7372 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7374 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7375 for TARGET_PREFER_AVX256.
7376 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7378 2017-09-21 Jeff Law <law@redhat.com>
7380 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7381 Fix dump output if the only stack space is for pushed registers.
7383 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7385 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7388 2017-09-21 Martin Sebor <msebor@redhat.com>
7391 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7392 code (in C++) or code that triggers warnings.
7394 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7396 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7398 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7400 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7401 * sched-rgn.c: Ditto.
7402 * sel-sched-ir.c: Ditto.
7404 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7406 * toplev.h (set_random_seed): Adjust return type.
7407 * toplev.c (init_local_tick): Move eager initialization of random_seed
7408 to get_random_seed. Adjust comment.
7409 (init_random_seed): Inline to get_random_seed, delete.
7410 (get_random_seed): Initialize random_seed lazily.
7411 (set_random_seed): Do not return previous value.
7412 (print_switch_value): Do not call get_random_seed.
7414 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7416 * cgraph.c (delete_function_version): New, broken out from...
7417 (cgraph_node::delete_function_version): ...here. Rename to
7418 cgraph_node::delete_function_version_by_decl. Update all uses.
7419 (cgraph_node::remove): Call delete_function_version.
7421 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7424 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7425 VAR_DECLs to which addressable non-volatile parameters are mapped
7426 and for id->retvar after the return value assignment. Clear
7427 id->retval and id->retbnd after inlining.
7429 2017-09-21 Richard Biener <rguenther@suse.de>
7431 PR tree-optimization/82276
7432 PR tree-optimization/82244
7433 * tree-vrp.c (build_assert_expr_for): Set
7434 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7436 (remove_range_assertions): Revert earlier change.
7438 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7441 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7443 2017-09-21 Richard Biener <rguenther@suse.de>
7445 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7446 Restore valid IL after code generation errors.
7447 * graphite.c (graphite_transform_loops): Diagnose code
7448 generation issues as MSG_MISSED_OPTIMIZATION and continue
7449 with processing SCOPs.
7451 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7452 Alan Hayward <alan.hayward@arm.com>
7453 David Sherwood <david.sherwood@arm.com>
7455 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7456 rather than choosing between plus_constant and gen_rtx_<CODE>.
7457 * expr.c (emit_push_insn): Likewise.
7458 (expand_expr_real_2): Likewise.
7460 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7461 Alan Hayward <alan.hayward@arm.com>
7462 David Sherwood <david.sherwood@arm.com>
7464 * loop-unroll.c (split_iv): Call copy_rtx on the step.
7466 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7467 Alan Hayward <alan.hayward@arm.com>
7468 David Sherwood <david.sherwood@arm.com>
7470 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7471 calling tree_to_uhwi.
7473 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7474 Alan Hayward <alan.hayward@arm.com>
7475 David Sherwood <david.sherwood@arm.com>
7477 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7478 INTEGER_CST rather than a negative test for ADDR_EXPR.
7480 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7481 Alan Hayward <alan.hayward@arm.com>
7482 David Sherwood <david.sherwood@arm.com>
7484 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7485 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7487 2017-09-21 Richard Biener <rguenther@suse.de>
7489 PR tree-optimization/71351
7490 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7491 graphite_create_new_loop_guard): Remove, fold remaining parts
7493 (translate_isl_ast_node_for): ... here and simplify.
7495 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7498 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7499 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7500 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7501 alternative always use QI mode, for -Os imov (=R,R) alternative
7502 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7505 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7506 Jeff Law <law@redhat.com>
7508 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7509 (allocate_stack_space): New function, partially extracted from
7511 (s390_emit_prologue): Track offset to most recent stack probe.
7512 Code to allocate space moved into allocate_stack_space.
7513 Dump actions when no stack is allocated.
7514 (s390_prologue_plus_offset): New function.
7515 (s390_emit_stack_probe): Likewise.
7517 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7519 * common.opt (Wa, Wl, Wp, g, gz=): Add
7521 (gno-column-info): Remove.
7522 (gcolumn-info): Drop RejectNegative.
7524 (gno-record-gcc-switches): Remove.
7525 (grecord-gcc-switches): Drop RejectNegative.
7526 (gno-split-dwarf): Remove.
7527 (gsplit-dwarf): Drop RejectNegative.
7528 (gno-strict-dwarf): Remove.
7529 (gstrict-dwarf): Drop RejectNegative.
7530 * config/darwin.opt (gfull, gused): Add RejectNegative.
7531 * dwarf2out.c (gen_producer_string): Drop
7532 gno-record-gcc-switches handler.
7533 * optc-gen.awk: Add g to prefixes with negative forms.
7534 * opts-common.c (remapping_prefix_p): New.
7535 (find_opt): Check it.
7536 (generate_canonical_option): Test g prefix.
7537 (option_map): Add -gno- mapping.
7538 (add_misspelling_candidates): Check remapping_prefix_p.
7540 2017-09-20 Jeff Law <law@redhat.com>
7542 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7543 thinko in stack clash protection support.
7545 * explow.c (compute_stack_clash_protection_loop_data): Use
7546 CONST_INT_P instead of explicit test. Verify object is a
7547 CONST_INT_P before looking at INTVAL.
7548 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7549 instead of explicit test.
7551 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7554 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7555 address instead of to r1 and r11.
7557 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
7559 * config.gcc: Support "knm".
7560 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
7561 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7563 * config/i386/i386.c (m_KNM): Define.
7564 (processor_target_table): Add "knm".
7566 (ix86_option_override_internal): Add "knm".
7567 (ix86_issue_rate): Add PROCESSOR_KNM.
7568 (ix86_adjust_cost): Ditto.
7569 (ia32_multipass_dfa_lookahead): Ditto.
7570 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
7571 (fold_builtin_cpu): Add M_INTEL_KNM.
7572 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
7573 (processor_type): Add PROCESSOR_KNM.
7574 * config/i386/x86-tune.def: Add m_KNM.
7575 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
7577 2017-09-20 Richard Biener <rguenther@suse.de>
7579 PR tree-optimization/80213
7580 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
7581 are allowed in empty BBs as well.
7582 (canonicalize_loop_closed_ssa): Also look for other complex
7584 (scop_detection::get_sese): Include the loop-closed PHI block
7586 (scop_detection::merge_sese): Remove code adding extra blocks.
7587 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
7588 (build_scops): Assert the final returned scop is invalid.
7590 2017-09-20 Richard Biener <rguenther@suse.de>
7592 PR tree-optimization/82264
7593 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
7595 (vn_phi_lookup): Likewise.
7596 (vn_phi_insert): Likewise.
7598 2017-09-20 Jakub Jelinek <jakub@redhat.com>
7600 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
7601 that fits into uhwi or shwi, add DW_AT_const_value regardless
7602 of early_dwarf without going through RTL, using add_AT_unsigned
7605 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
7606 (DEBUG_LTO_ABBREV_SECTION): Likewise.
7607 (DEBUG_LTO_MACINFO_SECTION): Likewise.
7608 (DEBUG_MACRO_SECTION): Likewise.
7609 (DEBUG_LTO_MACRO_SECTION): Likewise.
7610 (DEBUG_STR_DWO_SECTION): Likewise.
7611 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
7612 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
7613 (DEBUG_LTO_DWO_LINE_SECTION): Define.
7614 (DEBUG_LTO_LINE_STR_SECTION): Define.
7615 (init_sections_and_labels): Initialize debug_line_str_section
7616 variable. Initialize debug_loc_section for -gdwarf-5 to
7617 DEBUG_LOCLISTS_SECTION. Formatting fixes.
7619 2017-09-20 Richard Biener <rguenther@suse.de>
7621 * graphite-sese-to-poly.c (extract_affine): Properly handle
7622 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
7624 2017-09-20 Richard Biener <rguenther@suse.de>
7626 PR tree-optimization/81373
7627 * graphite-scop-detection.c (build_cross_bb_scalars_def):
7628 Force SESE live-out defs to be handled even if they are
7631 2017-09-19 Jeff Law <law@redhat.com>
7633 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
7634 nothing for stack adjustments with REG_STACK_CHECK.
7635 * sched-deps.c (parse_add_or_inc): Reject insns with
7636 REG_STACK_CHECK from dependency breaking.
7637 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
7638 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
7639 * reg-notes.def (STACK_CHECK): New note.
7641 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
7642 (ix86_expand_prologue): Dump stack clash info as needed.
7643 Call ix86_adjust_stack_and_probe_stack_clash as needed.
7645 * function.c (dump_stack_clash_frame_info): New function.
7646 * function.h (dump_stack_clash_frame_info): Prototype.
7647 (enum stack_clash_probes): New enum.
7649 * config/alpha/alpha.c (alpha_expand_prologue): Also check
7650 flag_stack_clash_protection.
7651 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
7652 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
7653 (arm_frame_pointer_required): Likewise.
7654 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
7655 (ia64_expand_prologue): Likewise.
7656 * config/mips/mips.c (mips_expand_prologue): Likewise.
7657 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
7658 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7659 (sparc_flat_expand_prologue): Likewise.
7660 * config/spu/spu.c (spu_expand_prologue): Likewise.
7662 * explow.c: Include "params.h".
7663 (anti_adjust_stack_and_probe_stack_clash): New function.
7664 (get_stack_check_protect): Likewise.
7665 (compute_stack_clash_protection_loop_data): Likewise.
7666 (emit_stack_clash_protection_loop_start): Likewise.
7667 (emit_stack_clash_protection_loop_end): Likewise.
7668 (allocate_dynamic_stack_space): Use get_stack_check_protect.
7669 Use anti_adjust_stack_and_probe_stack_clash.
7670 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
7671 (emit_stack_clash_protection_loop_start): Likewise.
7672 (emit_stack_clash_protection_loop_end): Likewise.
7673 * rtl.h (get_stack_check_protect): Prototype.
7674 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
7675 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
7676 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
7678 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
7680 * doc/tm.texi: Rebuilt.
7681 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7682 get_stack_check_protect.
7683 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
7684 * config/arm/arm.c (arm_expand_prologue): Likewise.
7685 (arm_frame_pointer_required): Likewise.
7686 * config/i386/i386.c (ix86_expand_prologue): Likewise.
7687 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
7688 * config/mips/mips.c (mips_expand_prologue): Likewise.
7689 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
7690 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
7691 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7692 (sparc_flat_expand_prologue): Likewise.
7694 * common.opt (-fstack-clash-protection): New option.
7695 * flag-types.h (enum stack_check_type): Note difference between
7696 -fstack-check= and -fstack-clash-protection.
7697 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
7698 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
7699 * toplev.c (process_options): Issue warnings/errors for cases
7700 not handled with -fstack-clash-protection.
7701 * doc/invoke.texi (-fstack-clash-protection): Document new option.
7702 (-fstack-check): Note additional problem with -fstack-check=generic.
7703 Note that -fstack-check is primarily for Ada and refer users
7704 to -fstack-clash-protection for stack-clash-protection.
7705 Document new params for stack clash protection.
7707 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
7709 * config/i386/i386.c (ix86_split_long_move): Do not handle
7710 address used for LEA in a special way.
7712 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7714 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
7716 2017-09-19 Martin Sebor <msebor@redhat.com>
7719 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
7720 of incompatible types.
7722 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
7724 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7725 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
7726 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7727 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
7729 2017-09-19 Richard Biener <rguenther@suse.de>
7731 PR tree-optimization/82244
7732 * tree-vrp.c (remove_range_assertions): Do not propagate
7733 a constant to abnormals but replace the assert with a copy.
7735 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
7737 PR rtl-optimization/57878
7738 PR rtl-optimization/68988
7739 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7740 avoidance test involving non_reload_pseudos. Move frequency test
7741 below the general fragmentation avoidance test.
7743 2017-09-19 Richard Biener <rguenther@suse.de>
7745 PR tree-optimization/69728
7746 * graphite-sese-to-poly.c (schedule_error): New global.
7747 (add_loop_schedule): Handle empty domain by failing the
7749 (build_original_schedule): Handle schedule_error.
7751 2017-09-19 Richard Biener <rguenther@suse.de>
7753 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7754 Do not iterate to sibling loops but only to siblings of inner
7757 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7760 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7761 that the registers are different.
7763 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7765 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7766 to processor_model and "amdfam17h" to arch_names_table.
7767 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
7769 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7772 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
7774 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7775 Alan Hayward <alan.hayward@arm.com>
7776 David Sherwood <david.sherwood@arm.com>
7778 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
7780 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
7782 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
7783 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
7784 vect_schedule_slp_instance.
7785 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
7786 Update call to vect_slp_analyze_node_operations. Simplify return
7788 (vect_slp_analyze_bb_1): Update call accordingly.
7789 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
7790 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
7791 (vect_schedule_slp): Update call accordingly.
7793 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7794 Alan Hayward <alan.hayward@arm.com>
7795 David Sherwood <david.sherwood@arm.com>
7797 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
7798 with types that aren't in fact scalar.
7800 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7802 * tree-vect-slp.c (vect_record_max_nunits): New function,
7804 (vect_build_slp_tree_1): ...here.
7805 (vect_build_slp_tree_2): Call it for phis too.
7807 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7809 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
7810 to vect_get_vec_def_for_operand when getting the mask operand.
7812 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7813 Alan Hayward <alan.hayward@arm.com>
7814 David Sherwood <david.sherwood@arm.com>
7816 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7819 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7820 Alan Hayward <alan.hayward@arm.com>
7821 David Sherwood <david.sherwood@arm.com>
7823 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7824 calculation for vector booleans.
7826 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7827 Alan Hayward <alan.hayward@arm.com>
7828 David Sherwood <david.sherwood@arm.com>
7830 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7832 (vect_transform_stmt): ...here.
7833 (vect_analyze_stmt): Use it instead of calling
7834 vectorizable_live_operation directly.
7836 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7838 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7839 non-SIMT targets in acc vector loops.
7841 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7843 * configure.ac: Add arc and check if assembler supports gdwarf2.
7844 * configure: Regenerate.
7846 2017-09-18 Richard Biener <rguenther@suse.de>
7848 PR tree-optimization/82220
7849 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7850 epilogue niters from the min_profitable_iters compute.
7852 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7855 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7856 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7857 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7859 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7862 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7863 switching to a new text section.
7865 2017-09-18 Richard Biener <rguenther@suse.de>
7867 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7869 (build_alias_set): Reject aliases with no access function.
7871 2017-09-18 Richard Biener <rguenther@suse.de>
7873 PR tree-optimization/79622
7874 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7876 (build_cross_bb_scalars_use): Likewise.
7878 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7880 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7882 2017-09-18 Alan Modra <amodra@gmail.com>
7885 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7886 stack_pointer_rtx for count 0. Update comments. Break up
7887 large rtl expression.
7889 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7891 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7892 Increase to 20 bytes.
7893 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7894 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7895 or avx version of the stub.
7897 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7900 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7901 compute the minimum stack alignment. Also update preferred stack
7902 boundary for leaf functions.
7904 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7906 PR tree-optimization/82228
7907 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7910 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7912 * common/config/nds32/nds32-common.c
7913 (nds32_option_optimization_table): Refine formatting.
7914 (nds32_option_optimization_table): Use -fsched-pressure and
7915 -fomit-frame-pointer for specific optimization level.
7917 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7919 * config/nds32/nds32.c: Refine formatting and comments.
7920 * config/nds32/nds32.h: Likewise.
7921 * config/nds32/nds32.md: Likewise.
7922 * config/nds32/nds32-cost.c: Likewise.
7923 * config/nds32/nds32-isr.c: Likewise.
7924 * config/nds32/nds32-md-auxiliary.c: Likewise.
7925 * config/nds32/nds32-multiple.md: Likewise.
7926 * config/nds32/nds32-predicates.c: Likewise.
7928 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7929 Jakub Jelinek <jakub@redhat.com>
7931 Add support for -std=c++2a.
7932 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7934 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7936 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7939 * doc/extend.texi (Common Function Attributes): Add
7940 references to ARM, AArch64, and S/390 specific attributes.
7941 (Function Specific Option Pragmas): Add AArch64 and S/390
7942 to list of back ends that support the target pragma.
7944 2017-09-15 Nathan Sidwell <nathan@acm.org>
7946 * doc/standards.texi: Fix C++17 description. Update URLs for
7949 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7951 * common.opt (Wcast-align=strict): New warning option.
7952 * doc/invoke.texi: Document -Wcast-align=strict.
7954 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7956 * cgraph.h (cgraph_thunk_info): Add comments.
7957 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7958 assert for VIRTUAL_* arguments stricter.
7960 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7962 PR tree-optimization/71026
7963 * match.pd: Move RDIV patterns from fold-const.c
7964 * fold-const.c (distribute_real_division): Removed.
7965 (fold_binary_loc): Remove calls to distribute_real_divison.
7967 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7969 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7970 c++1z and gnu++1z as deprecated. Change other references to
7971 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7972 Change -Wc++1z-compat to -Wc++17-compat.
7973 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7974 * dwarf2out.c (highest_c_language): Handle C++17.
7975 (gen_compile_unit_die): Likewise.
7977 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7979 PR rtl-optimization/82192
7980 * combine.c (make_extraction): Don't look through non-paradoxical
7981 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7984 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7985 Alan Hayward <alan.hayward@arm.com>
7986 David Sherwood <david.sherwood@arm.com>
7988 * target.def (function_arg_offset): New hook.
7989 * targhooks.h (default_function_arg_offset): Declare.
7990 * targhooks.c (default_function_arg_offset): New function.
7991 * function.c (locate_and_pad_parm): Use
7992 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7993 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7994 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7995 * doc/tm.texi: Regenerate.
7996 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7997 * config/spu/spu.c (spu_function_arg_offset): New function.
7998 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7999 * system.h (FUNCTION_ARG_OFFSET): Poison.
8001 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8002 Alan Hayard <alan.hayward@arm.com>
8003 David Sherwood <david.sherwood@arm.com>
8005 * target.def (truly_noop_truncation): New hook.
8006 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8007 than TRULY_NOOP_TRUNCATION.
8008 * hooks.h (hook_bool_uint_uint_true): Declare.
8009 * hooks.c (hook_bool_uint_uint_true): New function.
8010 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8011 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8012 * doc/tm.texi: Regenerate.
8013 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8014 rather than TRULY_NOOP_TRUNCATION in comments.
8015 (simplify_comparison): Likewise.
8016 (record_truncated_value): Likewise.
8017 * expmed.c (extract_bit_field_1): Likewise.
8018 (extract_split_bit_field): Likewise.
8019 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8020 instead of TRULY_NOOP_TRUNCATION.
8021 * function.c (assign_parm_setup_block): Likewise.
8022 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8023 * rtlhooks.c: Include target.h.
8024 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8025 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8026 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8027 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8028 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8029 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8030 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8031 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8032 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8033 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8034 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8035 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8036 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8037 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8038 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8039 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8040 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8041 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8042 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8043 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8044 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8045 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8046 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8047 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8048 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8049 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8050 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8051 rather than TRULY_NOOP_TRUNCATION in comments.
8052 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8053 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8054 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8055 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8056 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8057 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8058 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8059 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8060 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8061 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8062 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8063 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8064 rather than TRULY_NOOP_TRUNCATION in comments.
8065 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8066 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8067 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8068 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8069 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8070 TRULY_NOOP_TRUNCATION condition.
8071 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8072 (TRULY_NOOP_TRUNCATION): Delete.
8073 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8074 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8075 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8076 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8077 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8078 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8079 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8080 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8081 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8082 rather than TRULY_NOOP_TRUNCATION in comments.
8083 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8084 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8085 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8086 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8087 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8088 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8090 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8093 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8094 (*cmp_ior): Likewise.
8095 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8096 (*ior_scc_scc_cmp): Likewise.
8097 (*and_scc_scc): Likewise.
8098 (*and_scc_scc_cmp): Likewise.
8100 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8101 Alan Hayard <alan.hayward@arm.com>
8102 David Sherwood <david.sherwood@arm.com>
8104 * target.def (can_change_mode_class): New hook.
8105 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8106 (hard_regno_nregs): Likewise.
8107 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8108 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8109 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8110 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8111 (LOAD_EXTEND_OP): Update accordingly.
8112 * doc/tm.texi: Regenerate.
8113 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8114 CANNOT_CHANGE_MODE_CLASS.
8115 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8116 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8117 * combine.c (simplify_set): Update accordingly.
8118 * emit-rtl.c (validate_subreg): Likewise.
8119 * recog.c (general_operand): Likewise.
8120 * regcprop.c (mode_change_ok): Likewise.
8121 * reload1.c (choose_reload_regs): Likewise.
8122 (inherit_piecemeal_p): Likewise.
8123 * rtlanal.c (simplify_subreg_regno): Likewise.
8124 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8125 instead of CANNOT_CHANGE_MODE_CLASS.
8126 (reload_cse_simplify_operands): Likewise.
8127 * reload.c (push_reload): Use targetm.can_change_mode_class
8128 instead of CANNOT_CHANGE_MODE_CLASS.
8129 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8130 REG_CANNOT_CHANGE_MODE_P.
8131 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8132 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8133 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8134 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8135 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8136 (arm_can_change_mode_class): New function.
8137 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8138 than CANNOT_CHANGE_MODE_CLASS in comments.
8139 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8140 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8141 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8142 (ix86_can_change_mode_class): ...this new function, inverting the
8143 sense of the return value.
8144 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8145 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8146 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8147 (ia64_can_change_mode_class): New function.
8148 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8149 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8150 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8151 (m32c_can_change_mode_class): ...this new function, inverting the
8152 sense of the return value.
8153 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8154 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8155 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8156 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8157 (mips_can_change_mode_class): ...this new function, inverting the
8158 sense of the return value.
8159 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8160 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8161 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8162 (msp430_can_change_mode_class): New function.
8163 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8164 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8165 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8166 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8167 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8168 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8169 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8170 (pa_cannot_change_mode_class): Replace with...
8171 (pa_can_change_mode_class): ...this new function, inverting the
8172 sense of the return value.
8173 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8174 than CANNOT_CHANGE_MODE_CLASS in comments.
8175 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8176 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8177 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8178 (pdp11_cannot_change_mode_class): Replace with...
8179 (pdp11_can_change_mode_class): ...this new function, inverting the
8180 sense of the return value.
8181 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8182 * config/powerpcspe/powerpcspe-protos.h
8183 (rs6000_cannot_change_mode_class_ptr): Delete.
8184 * config/powerpcspe/powerpcspe.c
8185 (rs6000_cannot_change_mode_class_ptr): Delete.
8186 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8187 (rs6000_option_override_internal): Assign to
8188 targetm.can_change_mode_class instead of
8189 rs6000_cannot_change_mode_class_ptr.
8190 (rs6000_cannot_change_mode_class): Replace with...
8191 (rs6000_can_change_mode_class): ...this new function, inverting the
8192 sense of the return value.
8193 (rs6000_debug_cannot_change_mode_class): Replace with...
8194 (rs6000_debug_can_change_mode_class): ...this new function.
8195 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8196 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8197 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8198 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8199 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8201 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8202 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8203 (rs6000_option_override_internal): Assign to
8204 targetm.can_change_mode_class instead of
8205 rs6000_cannot_change_mode_class_ptr.
8206 (rs6000_cannot_change_mode_class): Replace with...
8207 (rs6000_can_change_mode_class): ...this new function, inverting the
8208 sense of the return value.
8209 (rs6000_debug_cannot_change_mode_class): Replace with...
8210 (rs6000_debug_can_change_mode_class): ...this new function.
8211 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8212 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8213 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8214 (s390_can_change_mode_class): ...this new function, inverting the
8215 sense of the return value.
8216 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8217 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8218 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8219 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8220 (sh_cannot_change_mode_class): Replace with...
8221 (sh_can_change_mode_class): ...this new function, inverting the
8222 sense of the return value.
8223 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8224 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8225 (sparc_can_change_mode_class): New function.
8226 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8227 * config/spu/spu.c (spu_can_change_mode_class): New function.
8228 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8229 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8230 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8231 (visium_can_change_mode_class): New function.
8232 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8234 2017-09-15 Richard Biener <rguenther@suse.de>
8236 PR tree-optimization/82217
8237 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8238 but not undefined case.
8240 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8243 * postreload.c (reload_cse_simplify_operands): Skip
8244 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8246 2017-09-15 Richard Biener <rguenther@suse.de>
8248 PR tree-optimization/68823
8249 * graphite-scop-detection.c (build_alias_set): If we have a
8250 possible dependence check whether we can handle them by just
8251 looking at the DRs DR_ACCESS_FNs.
8252 (build_scops): If build_alias_set fails, fail the SCOP.
8254 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8256 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8257 to support float128 built-in functions that require the ISA 3.0
8259 (BU_FLOAT128_3_HW): Likewise.
8260 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8262 (FMAF128): Likewise.
8264 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8265 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8266 floating point instructions.
8267 (rs6000_invalid_builtin): Likewise.
8268 (rs6000_builtin_mask_names): Likewise.
8269 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8270 (RS6000_BTM_FLOAT128_HW): Likewise.
8271 (RS6000_BTM_COMMON): Likewise.
8272 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8274 * doc/extend.texi (RS/6000 built-in functions): Document the
8275 IEEE 128-bit floating point square root and fused multiply-add
8278 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8280 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8281 reg (r2) isn't in the set of registers defined in the prologue.
8283 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8284 Alan Hayward <alan.hayward@arm.com>
8285 David Sherwood <david.sherwood@arm.com>
8287 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8288 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8289 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8290 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8291 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8293 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8294 max_vectorization_factor.
8295 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8297 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8298 Alan Hayward <alan.hayward@arm.com>
8299 David Sherwood <david.sherwood@arm.com>
8301 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8302 (vect_worthwhile_without_simd_p): Declare.
8303 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8304 (vectorizable_reduction): Use it.
8305 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8306 (vectorizable_operation): Likewise.
8308 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8309 Alan Hayward <alan.hayward@arm.com>
8310 David Sherwood <david.sherwood@arm.com>
8312 * tree-vectorizer.h (vect_get_num_copies): New function.
8313 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8314 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8315 (vectorizable_induction): Likewise.
8316 (vectorizable_live_operation): Likewise.
8317 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8318 (vectorizable_bswap): Likewise.
8319 (vectorizable_call): Likewise.
8320 (vectorizable_conversion): Likewise.
8321 (vectorizable_assignment): Likewise.
8322 (vectorizable_shift): Likewise.
8323 (vectorizable_operation): Likewise.
8324 (vectorizable_store): Likewise.
8325 (vectorizable_load): Likewise.
8326 (vectorizable_condition): Likewise.
8327 (vectorizable_comparison): Likewise.
8328 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8330 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8331 Alan Hayward <alan.hayward@arm.com>
8332 David Sherwood <david.sherwood@arm.com>
8334 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8335 of vect_init_vector.
8337 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8338 Alan Hayward <alan.hayward@arm.com>
8339 David Sherwood <david.sherwood@arm.com>
8341 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8342 an inline wrapper that provides a location.
8343 (gimple_build_vector): Likewise.
8344 * gimple-fold.c (gimple_build_vector_from_val): New function.
8345 (gimple_build_vector): Likewise.
8346 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8347 functions to build the initial value. Always return a gimple value.
8348 (get_initial_defs_for_reduction): Likewise. Only compute
8350 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8351 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8352 (vectorizable_induction): Use gimple_build_vector rather than
8355 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8356 Alan Hayward <alan.hayward@arm.com>
8357 David Sherwood <david.sherwood@arm.com>
8359 * target.h (vec_perm_indices): New typedef.
8360 (auto_vec_perm_indices): Likewise.
8361 * optabs-query.h: Include target.h
8362 (can_vec_perm_p): Take a vec_perm_indices *.
8363 * optabs-query.c (can_vec_perm_p): Likewise.
8364 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8365 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8366 * tree-vect-generic.c (lower_vec_perm): Likewise.
8367 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8368 (vect_grouped_load_supported): Likewise.
8369 (vect_shift_permute_load_chain): Likewise.
8370 (vect_permute_store_chain): Use auto_vec_perm_indices.
8371 (vect_permute_load_chain): Likewise.
8372 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8373 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8374 Update uses of can_vec_perm_p.
8375 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8376 mode with a number of elements. Take a vec_perm_indices *.
8377 (vect_create_epilog_for_reduction): Update accordingly.
8378 Use auto_vec_perm_indices.
8379 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8380 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8381 (vect_transform_slp_perm_load): Likewise.
8382 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8383 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8384 (vect_gen_perm_mask_checked): Likewise.
8385 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8386 (vect_gen_perm_mask_checked): Likewise.
8387 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8388 (vectorizable_store): Likewise.
8389 (vectorizable_load): Likewise.
8390 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8391 (vectorizable_bswap): Likewise.
8393 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8394 Alan Hayward <alan.hayward@arm.com>
8395 David Sherwood <david.sherwood@arm.com>
8397 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8398 * tree.c (build_vector): Likewise.
8399 (build_vector_from_ctor): Update accordingly.
8400 (build_vector_from_val): Likewise.
8401 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8402 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8403 * tree-vect-generic.c (add_rshift): Likewise.
8404 (expand_vector_divmod): Likewise.
8405 (optimize_vector_constructor): Likewise.
8406 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8407 (vect_transform_slp_perm_load): Likewise.
8408 (vect_schedule_slp_instance): Likewise.
8409 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8410 (vectorizable_call): Likewise.
8411 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8412 * expmed.c (make_tree): Likewise.
8413 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8414 a vector passed to build_vector.
8415 (fold_convert_const): Likewise.
8416 (exact_inverse): Likewise.
8417 (fold_ternary_loc): Likewise.
8418 (fold_relational_const): Likewise.
8419 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8420 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8421 (const_unop): Likewise. Store the reduction accumulator in a
8422 variable rather than an array.
8423 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8424 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8425 the new vector, rather than constructing it after the input arrays.
8426 (native_interpret_vector): Use auto_vec<tree> when building
8427 a vector passed to build_vector. Add elements in order.
8428 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8429 auto_vec<tree> when building a vector passed to build_vector.
8430 (vect_create_epilog_for_reduction): Likewise.
8431 (vectorizable_induction): Likewise.
8432 (get_initial_def_for_reduction): Likewise. Fix indentation of
8434 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8436 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8437 passed to build_vector.
8439 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8440 Alan Hayward <alan.hayward@arm.com>
8441 David Sherwood <david.sherwood@arm.com>
8443 * tree-core.h (tree_base::u): Add an "nelts" field.
8444 (tree_vector): Use VECTOR_CST_NELTS as the length.
8445 * tree.c (tree_size): Likewise.
8446 (make_vector): Initialize VECTOR_CST_NELTS.
8447 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8448 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8449 TYPE_VECTOR_SUBPARTS.
8450 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8451 as the number of units, setting it from VECTOR_CST_NELTS.
8452 (const_vector_from_tree): Likewise.
8453 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8454 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8455 (fold_negate_expr_1): Likewise.
8456 (fold_convert_const): Likewise.
8457 (const_binop): Likewise. Differentiate the number of output and
8459 (const_unop): Likewise.
8460 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8461 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8462 in cases that did the opposite.
8464 2017-09-14 Richard Biener <rguenther@suse.de>
8466 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8469 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8471 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8473 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8476 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8477 if and where to split a bb, except for splitting before debug insn
8478 sequences followed by non-label real insn. Delete debug insns
8479 in between basic blocks.
8481 * combine.c (make_compound_operation_int): Formatting fixes.
8483 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8484 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8485 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8486 * config/sol2.h (LINK_EH_SPEC): Likewise.
8487 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8488 * config/s390/linux.h (LINK_SPEC): Likewise.
8489 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8490 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8491 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8492 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8493 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8494 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8495 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8496 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8498 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8500 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8502 (ENDFILE_LINUX_SPEC): Likewise.
8503 (LINK_EH_SPEC): Likewise.
8504 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8505 (LINK_OS_LINUX_SPEC32): Likewise.
8506 (LINK_OS_LINUX_SPEC64): Likewise.
8507 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8508 (LINK_OS_LINUX_SPEC): Likewise.
8510 2017-09-13 Martin Liska <mliska@suse.cz>
8513 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8514 CASE_HIGH is NULL_TREE.
8516 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8517 Alan Hayward <alan.hayward@arm.com>
8518 David Sherwood <david.sherwood@arm.com>
8520 * target.def (secondary_memory_needed): New hook.
8521 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8522 instead of SECONDARY_MEMORY_NEEDED.
8523 (secondary_memory_needed_mode): Likewise.
8524 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8525 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8526 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8527 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8528 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8529 * doc/tm.texi: Regenerate.
8530 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8531 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8532 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8533 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8534 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8535 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8536 mode argument first and change the reg_class arguments to reg_class_t.
8537 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8538 Make static. Update the call to inline_secondary_memory_needed.
8539 (ix86_register_move_cost): Update the call to
8540 inline_secondary_memory_needed.
8541 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8542 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8544 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8545 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8547 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8548 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8549 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8550 and match hook interface. Add comment from mips.h.
8551 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8552 * config/mmix/mmix.md (truncdfsf2): Refer to
8553 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8555 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
8556 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
8557 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8558 (pa_secondary_memory_needed): New function.
8559 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
8560 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
8561 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8562 (pdp11_secondary_memory_needed): Make static and match hook interface.
8563 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
8564 * config/powerpcspe/powerpcspe-protos.h
8565 (rs6000_secondary_memory_needed_ptr): Delete.
8566 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
8568 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8569 (rs6000_option_override_internal): Assign to
8570 targetm.secondary_memory_needed rather than
8571 rs6000_secondary_memory_needed_ptr.
8572 (rs6000_secondary_memory_needed): Match hook interface.
8573 (rs6000_debug_secondary_memory_needed): Likewise.
8574 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
8575 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
8576 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
8577 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8578 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
8579 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
8581 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
8582 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8583 (rs6000_option_override_internal): Assign to
8584 targetm.secondary_memory_needed rather than
8585 rs6000_secondary_memory_needed_ptr.
8586 (rs6000_secondary_memory_needed): Match hook interface.
8587 (rs6000_debug_secondary_memory_needed): Likewise.
8588 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
8589 * config/s390/s390.c (s390_secondary_memory_needed): New function.
8590 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8591 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
8592 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8593 (sparc_secondary_memory_needed): New function.
8594 * lra-constraints.c (check_and_process_move): Refer to
8595 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8597 (curr_insn_transform): Likewise.
8598 (process_alt_operands): Use targetm.secondary_memory_needed
8599 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8600 (check_secondary_memory_needed_p): Likewise.
8601 (choose_split_class): Likewise.
8602 * reload.c: Unconditionally include code that was previously
8603 conditional on SECONDARY_MEMORY_NEEDED.
8604 (push_secondary_reload): Use targetm.secondary_memory_needed
8605 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8606 (push_reload): Likewise.
8607 * reload1.c: Unconditionally include code that was previously
8608 conditional on SECONDARY_MEMORY_NEEDED.
8609 (choose_reload_regs): Use targetm.secondary_memory_needed
8610 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8611 (gen_reload): Likewise.
8612 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
8614 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8615 Alan Hayward <alan.hayward@arm.com>
8616 David Sherwood <david.sherwood@arm.com>
8618 * target.def (secondary_memory_needed_mode): New hook:
8619 * targhooks.c (default_secondary_memory_needed_mode): Declare.
8620 * targhooks.h (default_secondary_memory_needed_mode): New function.
8621 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
8622 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
8623 * doc/tm.texi: Regenerate.
8624 * lra-constraints.c (check_and_process_move): Use
8625 targetm.secondary_memory_needed_mode instead of
8626 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
8627 (curr_insn_transform): Likewise.
8628 * reload.c (get_secondary_mem): Likewise.
8629 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8630 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
8632 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8633 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8634 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
8635 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8636 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
8638 * config/powerpcspe/powerpcspe-protos.h
8639 (rs6000_secondary_memory_needed_mode): Delete.
8640 * config/powerpcspe/powerpcspe.c
8641 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8642 (rs6000_secondary_memory_needed_mode): Make static.
8643 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8644 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
8646 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8648 (rs6000_secondary_memory_needed_mode): Make static.
8649 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8650 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
8651 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8652 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8653 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8655 (sparc_secondary_memory_needed_mode): New function.
8656 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
8658 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
8660 * config/aarch64/constraints.md (Umq): New constraint.
8661 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
8663 (mov<mode>): Update condition.
8665 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8667 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
8668 when bitposition is the same.
8670 2017-09-13 Richard Biener <rguenther@suse.de>
8672 * dwarf2out.c (output_die_symbol): Remove.
8673 (output_die): Do not output a DIEs symbol.
8675 2017-09-13 Richard Biener <rguenther@suse.de>
8678 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
8679 default-def to avoid breaking iterator update with the weird
8680 interaction with cgraph_update_edges_for_call_stmt_node.
8682 2017-09-13 Richard Biener <rguenther@suse.de>
8684 * tree-cfg.c (verify_gimple_assign_binary): Add verification
8685 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
8686 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
8687 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
8689 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
8691 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
8692 Disable pc relative literal load irrespective of
8693 TARGET_FIX_ERR_A53_84341 for default.
8695 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
8697 * config/sparc/sparc.c (output_return): Output the source location of
8698 the insn in the delay slot, if any.
8699 (output_sibcall): Likewise.
8701 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
8704 * common.opt (-static-pie): New alias.
8705 (shared): Negate static-pie.
8706 (-no-pie): Update help text.
8708 (static-pie): New option.
8709 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
8710 -static-pie support.
8711 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
8712 (LINK_EH_SPEC): Likewise.
8713 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8714 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8715 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8716 * gcc.c (LINK_COMMAND_SPEC): Likewise.
8717 (init_gcc_specs): Likewise.
8718 (init_spec): Likewise.
8719 (display_help): Update help message for -pie.
8720 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
8723 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
8725 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
8726 (movdi_aarch64): Likewise.
8727 (movti_aarch64): Likewise.
8729 2017-09-12 Simon Wright <simon@pushface.org>
8732 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
8733 calculation of the minor version, always output as 0.
8735 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8738 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8739 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8740 on it early, rather than manual conversion late. For
8741 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8742 instead of performing manual conversion.
8744 2017-09-12 Carl Love <cel@us.ibm.com>
8746 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8747 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8749 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8750 VMULOSW): Add definitions.
8751 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8752 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8753 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8754 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8755 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8757 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8759 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8761 (movti_aarch64): Likewise.
8762 (movdf_aarch64): Likewise.
8763 (movtf_aarch64): Likewise.
8764 (load_pairdi): Likewise.
8765 (store_pairdi): Likewise.
8766 (load_pairdf): Likewise.
8767 (store_pairdf): Likewise.
8768 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
8769 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
8770 (ldr_got_small_<mode>): Likewise.
8771 (ldr_got_small_28k_<mode>): Likewise.
8772 (ldr_got_tiny): Likewise.
8773 * config/aarch64/iterators.md (ldst_sz): New.
8774 (ldpstp_sz): Likewise.
8775 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
8777 (thunderx_load): Split load_8 to load_16.
8778 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
8780 (thunderx2t99_storepair_basic): Split store_8 to store_16.
8781 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
8782 (xgene1_store_pair): Split store_8 to store_16.
8783 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
8784 (falkor_st_0_st_sd): Split store_8 to store_16.
8786 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8788 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
8789 and store1/2/3/4 to store_4/8/12/16.
8790 * config/aarch64/aarch64.md: Update for rename.
8791 * config/arm/arm.md: Likewise.: Likewise.
8792 * config/arm/arm.c: Likewise.
8793 * config/arm/thumb1.md: Likewise.
8794 * config/arm/thumb2.md: Likewise.
8795 * config/arm/vfp.md: Likewise.
8796 * config/arm/arm-generic.md: Likewise.
8797 * config/arm/arm1020e.md: Likewise.
8798 * config/arm/arm1026ejs.md: Likewise.
8799 * config/arm/arm1136jfs.md: Likewise.
8800 * config/arm/arm926ejs.md: Likewise.
8801 * config/arm/cortex-a15.md: Likewise.
8802 * config/arm/cortex-a17.md: Likewise.
8803 * config/arm/cortex-a5.md: Likewise.
8804 * config/arm/cortex-a53.md: Likewise.
8805 * config/arm/cortex-a57.md: Likewise.
8806 * config/arm/cortex-a7.md: Likewise.
8807 * config/arm/cortex-a8.md: Likewise.
8808 * config/arm/cortex-a9.md: Likewise.
8809 * config/arm/cortex-m4.md: Likewise.
8810 * config/arm/cortex-m7.md: Likewise.
8811 * config/arm/cortex-r4.md: Likewise.
8812 * config/arm/exynos-m1.md: Likewise.
8813 * config/arm/fa526.md: Likewise.
8814 * config/arm/fa606te.md: Likewise.
8815 * config/arm/fa626te.md: Likewise.
8816 * config/arm/fa726te.md: Likewise.
8817 * config/arm/fmp626.md: Likewise.
8818 * config/arm/iwmmxt.md: Likewise.
8819 * config/arm/ldmstm.md: Likewise.
8820 * config/arm/marvell-pj4.md: Likewise.
8821 * config/arm/xgene1.md: Likewise.
8822 * config/aarch64/thunderx.md: Likewise.
8823 * config/aarch64/thunderx2t99.md: Likewise.
8824 * config/aarch64/falkor.md: Likewise.
8826 2017-09-12 Martin Liska <mliska@suse.cz>
8828 * attribs.c (private_lookup_attribute): New function.
8829 * attribs.h (private_lookup_attribute): Declared here.
8830 (lookup_attribute): Called from this place.
8832 2017-09-12 Richard Biener <rguenther@suse.de>
8834 PR tree-optimization/82157
8835 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8836 stmts with side-effects.
8838 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8839 Alan Hayward <alan.hayward@arm.com>
8840 David Sherwood <david.sherwood@arm.com>
8842 * target.def (hard_regno_nregs): New hook.
8843 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8844 * targhooks.h (default_hard_regno_nregs): Declare.
8845 * targhooks.c (default_hard_regno_nregs): New function.
8846 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8847 (TARGET_HARD_REGNO_NREGS): ...this hook.
8848 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8849 (CLASS_MAX_NREGS): Likewise.
8850 * doc/tm.texi: Regenerate.
8851 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8852 instead of HARD_REGNO_NREGS.
8853 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8854 HARD_REGNO_NREGS in the comment.
8855 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8856 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8857 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8858 Return an unsigned int.
8859 (TARGET_HARD_REGNO_NREGS): Redefine.
8860 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8861 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8862 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8863 (arc_hard_regno_nregs): New function.
8864 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8865 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8866 (arm_hard_regno_nregs): New function.
8867 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8868 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8869 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8870 (TARGET_HARD_REGNO_NREGS): Redefine.
8871 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8872 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8873 (HARD_REGNO_NREGS): Delete.
8874 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8875 (cr16_hard_regno_nregs): New function.
8876 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8877 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8878 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8879 (cris_hard_regno_nregs): New function.
8880 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8881 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8882 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8883 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8884 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8885 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8886 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8887 (frv_hard_regno_nregs): Make static. Take and return an
8889 (frv_class_max_nregs): Remove outdated copy of documentation.
8890 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8891 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8892 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8893 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8894 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8895 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8896 (TARGET_HARD_REGNO_NREGS): Redefine.
8897 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8898 (CLASS_MAX_NREGS): Update comment.
8899 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8900 (ia64_hard_regno_nregs): New function.
8901 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8902 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8903 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8904 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8905 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8907 (m32c_hard_regno_nregs): Likewise. Make static.
8908 (TARGET_HARD_REGNO_NREGS): Redefine.
8909 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8910 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8911 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8912 (m68k_hard_regno_nregs): New function.
8913 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8914 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8915 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8916 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8917 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8918 Take and return an unsigned int.
8919 (TARGET_HARD_REGNO_NREGS): Redefine.
8920 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8921 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8922 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8923 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8924 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8925 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8926 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8927 (msp430_hard_regno_nregs): Make static. Take and return an
8929 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8930 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8931 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8932 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8933 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8934 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8935 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8936 (TARGET_HARD_REGNO_NREGS): Redefine.
8937 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8938 (PA_HARD_REGNO_NREGS): ...this.
8939 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8940 (PA_HARD_REGNO_NREGS): ...this.
8941 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8942 (pa_hard_regno_nregs): New function.
8943 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8944 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8945 (pdp11_hard_regno_nregs): New function.
8946 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8947 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8948 (rs6000_hard_regno_nregs_hook): New function.
8949 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8950 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8951 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8952 Take and return an unsigned int. Move earlier in file.
8953 (TARGET_HARD_REGNO_NREGS): Redefine.
8954 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8955 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8956 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8957 (rl78_hard_regno_nregs): Make static. Take and return an
8959 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8960 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8961 (rs6000_hard_regno_nregs_hook): New function.
8962 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8963 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8964 (TARGET_HARD_REGNO_NREGS): Redefine.
8965 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8966 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8967 instead of HARD_REGNO_NREGS.
8968 (s390_hard_regno_nregs): New function.
8969 (s390_hard_regno_mode_ok): Add comment from s390.h.
8970 (TARGET_HARD_REGNO_NREGS): Redefine.
8971 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8972 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8973 (sh_hard_regno_nregs): New function.
8974 (sh_pass_in_reg_p): Use it.
8975 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8976 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8977 (sparc_hard_regno_nregs): New function.
8978 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8979 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8980 (spu_function_arg_advance): Use it, supplying a valid register number.
8981 (TARGET_HARD_REGNO_NREGS): Redefine.
8982 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8983 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8984 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8985 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8986 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8987 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8988 (CLASS_MAX_NREGS): Remove copy of old documentation.
8989 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8990 (visium_hard_regno_nregs): New function.
8991 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8992 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8993 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8994 xtensa_hard_regno_nregs): New function.
8995 * system.h (HARD_REGNO_NREGS): Poison.
8997 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8999 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9000 hard_regno_nregs instead of HARD_REGNO_NREGS.
9001 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9002 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9003 (c6x_expand_epilogue): Likewise.
9004 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9005 (frv_read_iacc_argument): Likewise.
9006 * config/sh/sh.c: Include regs.h.
9007 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9008 (regs_used): Likewise.
9009 (output_stack_adjust): Likewise.
9010 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9011 * expmed.c: Include regs.h.
9012 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9013 * ree.c: Include regs.h.
9014 (combine_reaching_defs): Use hard_regno_nregs instead of
9016 (add_removable_extension): Likewise.
9018 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9020 * regs.h (hard_regno_nregs): Turn into a function.
9021 (end_hard_regno): Update accordingly.
9022 * caller-save.c (setup_save_areas): Likewise.
9023 (save_call_clobbered_regs): Likewise.
9024 (replace_reg_with_saved_mem): Likewise.
9025 (insert_restore): Likewise.
9026 (insert_save): Likewise.
9027 * combine.c (can_change_dest_mode): Likewise.
9028 (move_deaths): Likewise.
9029 (distribute_notes): Likewise.
9030 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9031 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9032 (rs6000_split_multireg_move): Likewise.
9033 (rs6000_register_move_cost): Likewise.
9034 (rs6000_memory_move_cost): Likewise.
9035 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9036 (rs6000_split_multireg_move): Likewise.
9037 (rs6000_register_move_cost): Likewise.
9038 (rs6000_memory_move_cost): Likewise.
9039 * cselib.c (cselib_reset_table): Likewise.
9040 (cselib_lookup_1): Likewise.
9041 * emit-rtl.c (set_mode_and_regno): Likewise.
9042 * function.c (aggregate_value_p): Likewise.
9043 * ira-color.c (setup_profitable_hard_regs): Likewise.
9044 (check_hard_reg_p): Likewise.
9045 (calculate_saved_nregs): Likewise.
9046 (assign_hard_reg): Likewise.
9047 (improve_allocation): Likewise.
9048 (calculate_spill_cost): Likewise.
9049 * ira-emit.c (modify_move_list): Likewise.
9050 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9051 (ira_hard_reg_in_set_p): Likewise.
9052 * ira.c (setup_reg_mode_hard_regset): Likewise.
9053 (clarify_prohibited_class_mode_regs): Likewise.
9054 (check_allocation): Likewise.
9055 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9056 (lra_setup_reg_renumber): Likewise.
9057 (setup_try_hard_regno_pseudos): Likewise.
9058 (spill_for): Likewise.
9059 (assign_hard_regno): Likewise.
9060 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9061 * lra-constraints.c (in_class_p): Likewise.
9062 (lra_constraint_offset): Likewise.
9063 (simplify_operand_subreg): Likewise.
9064 (lra_constraints): Likewise.
9065 (split_reg): Likewise.
9066 (split_if_necessary): Likewise.
9067 (invariant_p): Likewise.
9068 (inherit_in_ebb): Likewise.
9069 * lra-lives.c (process_bb_lives): Likewise.
9070 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9071 (get_hard_regs): Likewise.
9072 (do_remat): Likewise.
9073 * lra-spills.c (assign_spill_hard_regs): Likewise.
9074 * mode-switching.c (create_pre_exit): Likewise.
9075 * postreload.c (reload_combine_recognize_pattern): Likewise.
9076 * recog.c (peep2_find_free_register): Likewise.
9077 * regcprop.c (kill_value_regno): Likewise.
9078 (set_value_regno): Likewise.
9079 (copy_value): Likewise.
9080 (maybe_mode_change): Likewise.
9081 (find_oldest_value_reg): Likewise.
9082 (copyprop_hardreg_forward_1): Likewise.
9083 * regrename.c (check_new_reg_p): Likewise.
9084 (regrename_do_replace): Likewise.
9085 * reload.c (push_reload): Likewise.
9086 (combine_reloads): Likewise.
9087 (find_dummy_reload): Likewise.
9088 (operands_match_p): Likewise.
9089 (find_reloads): Likewise.
9090 (find_equiv_reg): Likewise.
9091 (reload_adjust_reg_for_mode): Likewise.
9092 * reload1.c (count_pseudo): Likewise.
9093 (count_spilled_pseudo): Likewise.
9094 (find_reg): Likewise.
9095 (clear_reload_reg_in_use): Likewise.
9096 (free_for_value_p): Likewise.
9097 (allocate_reload_reg): Likewise.
9098 (choose_reload_regs): Likewise.
9099 (reload_adjust_reg_for_temp): Likewise.
9100 (emit_reload_insns): Likewise.
9101 (delete_output_reload): Likewise.
9102 * rtlanal.c (subreg_get_info): Likewise.
9103 * sched-deps.c (sched_analyze_reg): Likewise.
9104 * sel-sched.c (init_regs_for_mode): Likewise.
9105 (mark_unavailable_hard_regs): Likewise.
9106 (choose_best_reg_1): Likewise.
9107 (verify_target_availability): Likewise.
9108 * valtrack.c (dead_debug_insert_temp): Likewise.
9109 * var-tracking.c (track_loc_p): Likewise.
9110 (emit_note_insn_var_location): Likewise.
9111 * varasm.c (make_decl_rtl): Likewise.
9112 * reginfo.c (choose_hard_reg_mode): Likewise.
9113 (init_reg_modes_target): Refer directly to
9114 this_target_regs->x_hard_regno_nregs.
9116 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9118 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9119 instead of hard_regno_nregs.
9121 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9123 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9124 end_hard_regno instead of hard_regno_nregs.
9125 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9126 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9127 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9128 * ira-color.c (improve_allocation): Likewise.
9129 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9130 * lra-lives.c (mark_regno_live): Likewise.
9131 (mark_regno_dead): Likewise.
9132 * lra-remat.c (operand_to_remat): Likewise.
9133 * lra.c (collect_non_operand_hard_regs): Likewise.
9134 * postreload.c (reload_combine_note_store): Likewise.
9135 (move2add_valid_value_p): Likewise.
9136 * reload.c (regno_clobbered_p): Likewise.
9138 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9140 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9142 * config/v850/v850.c (v850_reorg): Likewise.
9143 * reload.c (refers_to_regno_for_reload_p): Likewise.
9144 (find_equiv_reg): Likewise.
9145 * reload1.c (reload_reg_reaches_end_p): Likewise.
9147 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9149 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9151 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9152 * config/arm/arm.c (output_move_neon): Likewise.
9153 (arm_attr_length_move_neon): Likewise.
9154 (neon_split_vcombine): Likewise.
9155 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9156 (c6x_mark_reg_written): Likewise.
9157 (c6x_dwarf_register_span): Likewise.
9158 * config/i386/i386.c (ix86_save_reg): Likewise.
9159 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9160 (rws_access_reg): Likewise.
9161 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9162 * mode-switching.c (create_pre_exit): Likewise.
9163 * ree.c (combine_reaching_defs): Likewise.
9164 (add_removable_extension): Likewise.
9165 * regcprop.c (find_oldest_value_reg): Likewise.
9166 (copyprop_hardreg_forward_1): Likewise.
9167 * reload.c (reload_inner_reg_of_subreg): Likewise.
9168 (push_reload): Likewise.
9169 (combine_reloads): Likewise.
9170 (find_dummy_reload): Likewise.
9171 (reload_adjust_reg_for_mode): Likewise.
9172 * reload1.c (find_reload_regs): Likewise.
9173 (forget_old_reloads_1): Likewise.
9174 (reload_reg_free_for_value_p): Likewise.
9175 (reload_adjust_reg_for_temp): Likewise.
9176 (emit_reload_insns): Likewise.
9177 (delete_output_reload): Likewise.
9178 * sel-sched.c (choose_best_reg_1): Likewise.
9179 (choose_best_pseudo_reg): Likewise.
9181 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9182 Alan Hayward <alan.hayward@arm.com>
9183 David Sherwood <david.sherwood@arm.com>
9185 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9186 * target.def (slow_unaligned_access): New hook.
9187 * targhooks.h (default_slow_unaligned_access): Declare.
9188 * targhooks.c (default_slow_unaligned_access): New function.
9189 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9190 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9191 * doc/tm.texi: Regenerate.
9192 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9193 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9194 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9196 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9197 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9199 (rs6000_slow_unaligned_access): New function.
9200 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9201 (expand_block_compare): Likewise.
9202 (expand_strn_compare): Likewise.
9203 (rs6000_rtx_costs): Likewise.
9204 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9205 (riscv_slow_unaligned_access): Likewise.
9206 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9207 (riscv_slow_unaligned_access_p): ...this and make static.
9208 (riscv_option_override): Update accordingly.
9209 (riscv_slow_unaligned_access): New function.
9210 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9211 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9212 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9213 (rs6000_slow_unaligned_access): New function.
9214 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9215 (rs6000_rtx_costs): Likewise.
9216 * config/rs6000/rs6000-string.c (expand_block_compare)
9217 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9218 of SLOW_UNALIGNED_ACCESS.
9219 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9220 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9221 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9222 of SLOW_UNALIGNED_ACCESS.
9223 * expmed.c (simple_mem_bitfield_p): Likewise.
9224 * expr.c (alignment_for_piecewise_move): Likewise.
9225 (emit_group_load_1): Likewise.
9226 (emit_group_store): Likewise.
9227 (copy_blkmode_from_reg): Likewise.
9228 (emit_push_insn): Likewise.
9229 (expand_assignment): Likewise.
9230 (store_field): Likewise.
9231 (expand_expr_real_1): Likewise.
9232 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9233 * lra-constraints.c (simplify_operand_subreg): Likewise.
9234 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9235 * gimple-ssa-store-merging.c: Likewise in block comment at start
9237 * tree-ssa-strlen.c: Include target.h.
9238 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9239 of SLOW_UNALIGNED_ACCESS.
9240 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9242 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9244 PR rtl-optimization/82185
9245 * expmed.c (emit_store_flag_int): Only test tem if it has been
9248 2017-09-12 Richard Biener <rguenther@suse.de>
9251 * match.pd ((FTYPE) N CMP CST): Fix typo.
9253 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9255 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9257 (mips_near_type_p): Add 'short_call' attribute as a synonym
9259 * doc/extend.texi (short_call): Document new function attribute.
9261 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9264 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9265 assertion check that in the condition.
9266 (get_atomic_generic_size): Likewise. Before testing if parameter
9267 has pointer type, if it has array type, call for C++
9268 default_conversion to perform array-to-pointer conversion.
9270 2017-09-12 Richard Biener <rguenther@suse.de>
9272 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9273 for operations we cannot scalarize.
9275 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9277 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9278 vectors heap vectors. Clean up comments.
9279 Make visited_bbs a reference.
9280 (profitable_jump_thread_path): Make GC
9281 vectors heap vectors. Clean up comments.
9283 (convert_and_register_jump_thread_path): Make GC vectors heap
9285 (check_subpath_and_update_thread_path): Same. Clean up comments.
9286 Make visited_bbs a reference.
9287 (handle_phi): Abstract common code to to
9288 register_jump_thread_path_if_profitable.
9289 Rename VAR_BB to DEF_BB.
9291 Make GC vectors heap vectors.
9292 Make visited_bbs a reference.
9293 (handle_assignment): Same.
9294 (register_jump_thread_path_if_profitable): New.
9295 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9297 Make GC vectors heap vectors. Clean up comments.
9298 Make visited_bbs a reference.
9299 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9300 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9303 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9306 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9307 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9309 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9311 Revert r251800 and r251799.
9313 2017-09-11 Martin Jambor <mjambor@suse.cz>
9316 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9317 arguments in advance.
9318 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9319 use it to find predecessor edges.
9320 (naive_outof_ssa): Collect vector of predecessors.
9322 2017-09-08 Jason Merrill <jason@redhat.com>
9324 PR c++/70029 - ICE with ref-qualifier and -flto
9325 * langhooks.h (struct lang_hooks_for_types): Add
9326 copy_lang_qualifiers.
9327 * attribs.c (build_type_attribute_qual_variant): Use it.
9328 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9330 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9331 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9333 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9336 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9337 (*mulsi3_sp64): New instruction.
9338 (mulsi3): New expander.
9340 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9342 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9344 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9346 * sancov.c: Include memmodel.h.
9348 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9351 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9354 2017-09-07 Carl Love <cel@us.ibm.com>
9356 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9357 the sldi instruction.
9359 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9361 * sancov.c: Include tm_p.h.
9363 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9366 * output.h (switch_to_other_text_partition): New declaration.
9367 * varasm.c (switch_to_other_text_partition): New function.
9368 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9369 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9370 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9371 to the other text partition before emitting LCL label and switch back
9372 after emitting the word after it.
9374 2017-09-07 Richard Biener <rguenther@suse.de>
9376 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9377 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9378 critical edges here, after loop init.
9379 (pass_data_pre): Remove PROP_no_crit_edges flags.
9380 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9381 for valueization of call args to avoid leaking VN_TOP.
9382 (visit_use): Assert we do not visit default defs.
9383 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9384 Use error_mark_node to more easily detect leaking VN_TOP.
9385 All default-defs are varying, not VN_TOP. Mark them visited.
9386 (run_scc_vn): Make code match comment.
9388 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9390 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9391 OPTION_MASK_FLOAT128_KEYWORD.
9392 (POWERPC_MASKS): Likewise.
9393 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9394 support for the -mfloat128-type option, and make -mfloat128
9395 default on PowerPC Linux systems. Define or undefine
9396 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9397 Define __float128 to be __ieee128 if IEEE 128-bit support is
9398 enabled, or undefine it.
9399 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9400 Delete defining __FLOAT128_TYPE__.
9401 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9402 -mfloat128-type option and make -mfloat128 default on PowerPC
9404 (TARGET_FLOAT128_TYPE): Likewise.
9405 (-mfloat128-type): Likewise.
9406 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9407 Delete the -mfloat128-type option and make -mfloat128 default on
9408 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9409 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9410 128-bit floating point is enabled. Change tests from using
9411 -mfloat128-type to -mfloat128.
9412 (rs6000_mangle_type): Use the correct mangling for the __float128
9413 type even if normal long double is restricted to 64-bits.
9414 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9416 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9417 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9418 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9420 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9421 documentation for -mfloat128.
9423 2017-09-06 Olivier Hainque <hainque@adacore.com>
9425 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9427 2017-09-06 Wish Wu <wishwu007@gmail.com>
9428 Jakub Jelinek <jakub@redhat.com>
9430 * asan.c (initialize_sanitizer_builtins): Add
9431 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9432 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9433 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9434 BT_FN_VOID_UINT64_PTR variables.
9435 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9436 (BT_FN_VOID_UINT16_UINT16): Likewise.
9437 (BT_FN_VOID_UINT32_UINT32): Likewise.
9438 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9439 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9440 (BT_FN_VOID_UINT64_PTR): Likewise.
9441 * common.opt (flag_sanitize_coverage): New variable.
9442 (fsanitize-coverage=trace-pc): Remove.
9443 (fsanitize-coverage=): Add.
9444 * flag-types.h (enum sanitize_coverage_code): New enum.
9445 * fold-const.c (fold_range_test): Disable non-short-circuit
9446 optimization if flag_sanitize_coverage.
9447 (fold_truth_andor): Likewise.
9448 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9449 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9450 (coverage_sanitizer_opts): New array.
9451 (get_closest_sanitizer_option): Add OPTS argument, handle also
9452 OPT_fsanitize_coverage_.
9453 (parse_sanitizer_options): Adjusted to also handle
9454 OPT_fsanitize_coverage_.
9455 (common_handle_option): Add OPT_fsanitize_coverage_.
9456 * sancov.c (instrument_comparison, instrument_switch): New function.
9457 (sancov_pass): Add trace-cmp support.
9458 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9459 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9460 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9461 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9462 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9463 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9464 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9465 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9466 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9467 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9469 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9471 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9472 error. Only quit immediately if parsing is complete.
9473 (BEGIN): Initialize fatal_err and parse_done.
9474 (begin fpu, end fpu): Check number of arguments.
9475 (begin arch, end arch): Likewise.
9476 (begin cpu, end cpu): Likewise.
9477 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9478 (optalias): Likewise.
9480 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9482 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9483 * config/arm/arm-isa.h: Delete. Move definitions to ...
9484 * arm-cpus.in: ... here. Use new feature and fgroup values.
9485 * config/arm/arm.c (arm_option_override): Use lower case for feature
9487 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9488 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9489 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9491 (print_isa_bits_for): New function.
9492 (gen_isa): New function.
9493 (gen_comm_data): Use print_isa_bits_for.
9494 (define feature): New keyword.
9495 (define fgroup): New keyword.
9496 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9497 (arm-isa.h): Add rule to generate file.
9498 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9499 case for feature bit names.
9501 2017-09-06 Richard Biener <rguenther@suse.de>
9503 * tree-ssa-pre.c (NECESSARY): Remove.
9504 (create_expression_by_pieces): Do not touch pass-local flags.
9505 (insert_into_preds_of_block): Likewise.
9506 (do_pre_regular_insertion): Likewise.
9507 (eliminate_insert): Likewise.
9508 (eliminate_dom_walker::before_dom_children): Likewise.
9509 (fini_eliminate): Do not look at inserted_exprs.
9510 (mark_operand_necessary): Remove.
9511 (remove_dead_inserted_code): Replace with simple work-list
9512 algorithm based on inserted_exprs and SSA uses.
9513 (pass_pre::execute): Re-order fini_eliminate and
9514 remove_dead_inserted_code.
9516 2017-09-06 Olivier Hainque <hainque@adacore.com>
9518 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9519 for VxWorks 7. Adjust surrounding comments.
9521 2017-09-06 Richard Biener <rguenther@suse.de>
9523 * gimple-ssa-strength-reduction.c
9524 (find_candidates_dom_walker::before_dom_children): Also allow
9527 2017-09-06 Richard Biener <rguenther@suse.de>
9529 PR tree-optimization/82108
9530 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9531 for gap in the non-permutation SLP case.
9533 2017-09-06 Martin Jambor <mjambor@suse.cz>
9535 PR tree-optimization/82078
9536 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9537 add_access_to_work_queue...
9538 (build_accesses_from_assign): ...here.
9539 (propagate_all_subaccesses): Make sure racc is the group
9540 representative, if there is one.
9542 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9545 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9548 2017-09-06 Richard Biener <rguenther@suse.de>
9550 * gimple-ssa-strength-reduction.c
9551 (find_candidates_dom_walker::before_doom_children): Use a
9552 type and not a mode check.
9554 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9557 * config/arm/predicates.md (arm_general_adddi_operand): Create new
9559 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
9561 2017-09-05 Jeff Law <law@redhat.com>
9563 PR tree-optimization/64910
9564 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
9565 cases where we have 3 or more operands.
9567 2017-09-05 Jakub Jelinek <jakub@redhat.com>
9570 * omp-low.c (lower_omp_for): Recompute tree invariant if
9571 gimple_omp_for_initial/final is ADDR_EXPR.
9574 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
9575 into gimple val before gimplification fo the COND_EXPR.
9577 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
9579 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
9580 REGION_COPY argument.
9581 (thread_through_all_blocks): Remove unused argument to
9582 duplicate_thread_path.
9584 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9585 Alan Hayward <alan.hayward@arm.com>
9586 David Sherwood <david.sherwood@arm.com>
9588 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
9589 Take a scalar_mode rather than a machine_mode.
9590 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9591 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
9592 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9593 (aarch64_gen_adjusted_ldpstp): Likewise.
9594 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
9596 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9597 Alan Hayward <alan.hayward@arm.com>
9598 David Sherwood <david.sherwood@arm.com>
9600 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
9601 Take a scalar_int_mode instead of a machine_mode.
9602 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9603 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9604 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9605 (aarch64_simd_attr_length_rglist): Delete.
9606 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
9607 a scalar_int_mode instead of a machine_mode.
9608 (aarch64_add_offset): Likewise.
9609 (aarch64_internal_mov_immediate): Likewise
9610 (aarch64_add_constant_internal): Likewise.
9611 (aarch64_add_constant): Likewise.
9612 (aarch64_movw_imm): Likewise.
9613 (aarch64_rtx_arith_op_extract_p): Likewise.
9614 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9615 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9616 Remove assert that the mode isn't a vector.
9617 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9618 (aarch64_expand_mov_immediate): Update calls after above changes.
9619 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
9620 (aarch64_and_bitmask_imm): Check for scalar integer modes.
9621 (aarch64_move_imm): Likewise.
9622 (aarch64_can_const_movi_rtx_p): Likewise.
9623 (aarch64_strip_extend): Likewise.
9624 (aarch64_extr_rtx_p): Likewise.
9625 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
9626 a CONST_INT when the mode parameter is VOIDmode.
9627 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
9629 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9631 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
9632 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9633 (bitwise_type_for_mode): Update accordingly.
9635 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9637 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
9638 * stor-layout.c (mode_for_size_tree): Likewise.
9639 (mode_for_array): Update accordingly.
9640 (layout_decl): Likewise.
9641 (compute_record_mode): Likewise. Only set the mode once.
9643 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9645 * target.def (get_mask_mode): Change return type to opt_mode.
9647 * doc/tm.texi: Regenerate.
9648 * targhooks.h (default_get_mask_mode): Return an opt_mode.
9649 * targhooks.c (default_get_mask_mode): Likewise.
9650 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9651 * optabs-query.c (can_vec_mask_load_store_p): Update use of
9652 targetm.get_mask_mode.
9653 * tree.c (build_truth_vector_type): Likewise.
9655 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9657 * machmode.h (mode_for_vector): Return an opt_mode.
9658 * stor-layout.c (mode_for_vector): Likewise.
9659 (mode_for_int_vector): Update accordingly.
9660 (layout_type): Likewise.
9661 * config/i386/i386.c (emit_memmov): Likewise.
9662 (ix86_expand_set_or_movmem): Likewise.
9663 (ix86_expand_vector_init): Likewise.
9664 (ix86_get_mask_mode): Likewise.
9665 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
9667 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
9668 * expmed.c (extract_bit_field_1): Likewise.
9669 * expr.c (expand_expr_real_2): Likewise.
9670 * optabs-query.c (can_vec_perm_p): Likewise.
9671 (can_vec_mask_load_store_p): Likewise.
9672 * optabs.c (expand_vec_perm): Likewise.
9673 * targhooks.c (default_get_mask_mode): Likewise.
9674 * tree-vect-stmts.c (vectorizable_store): Likewise.
9675 (vectorizable_load): Likewise.
9676 (get_vectype_for_scalar_type_and_size): Likewise.
9678 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9680 * machmode.h (mode_for_int_vector): New function.
9681 * stor-layout.c (mode_for_int_vector): Likewise.
9682 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
9683 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
9684 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9685 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9686 (s390_expand_vcond): Likewise.
9688 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9690 * machmode.h (opt_machine_mode): New type.
9691 (opt_mode<T>): Allow construction from anything that can be
9693 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
9694 (mode_for_size): Return an opt_machine_mode.
9695 * stor-layout.c (mode_for_size): Likewise.
9696 (mode_for_size_tree): Update call accordingly.
9697 (bitwise_mode_for_mode): Likewise.
9698 (make_fract_type): Likewise.
9699 (make_accum_type): Likewise.
9700 * caller-save.c (replace_reg_with_saved_mem): Update call
9702 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9703 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9704 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9705 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9706 * expmed.c (extract_bit_field_1): Likewise.
9707 * reload.c (get_secondary_mem): Likewise.
9708 * varasm.c (assemble_integer): Likewise.
9709 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
9712 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9714 * machmode.h (decimal_float_mode_for_size): New function.
9715 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
9716 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
9717 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
9718 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
9719 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
9720 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
9722 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9724 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
9725 (get_builtin_sync_mode): Likewise.
9726 (expand_ifn_atomic_compare_exchange): Likewise.
9727 (expand_builtin_atomic_clear): Likewise.
9728 (expand_builtin_atomic_test_and_set): Likewise.
9729 (fold_builtin_atomic_always_lock_free): Likewise.
9730 * calls.c (compute_argument_addresses): Likewise.
9731 (emit_library_call_value_1): Likewise.
9732 (store_one_arg): Likewise.
9733 * combine.c (combine_instructions): Likewise.
9734 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
9735 * config/arm/arm.c (arm_function_value): Likewise.
9736 (aapcs_allocate_return_reg): Likewise.
9737 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
9738 * config/i386/i386.c (construct_container): Likewise.
9739 (ix86_gimplify_va_arg): Likewise.
9740 (ix86_expand_sse_cmp): Likewise.
9741 (emit_memmov): Likewise.
9742 (emit_memset): Likewise.
9743 (expand_small_movmem_or_setmem): Likewise.
9744 (ix86_expand_pextr): Likewise.
9745 (ix86_expand_pinsr): Likewise.
9746 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9747 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9749 * config/mips/mips.c (mips_function_value_1) Likewise.
9750 (mips_block_move_straight): Likewise.
9751 (mips_expand_ins_as_unaligned_store): Likewise.
9752 * config/powerpcspe/powerpcspe.c
9753 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9754 (rs6000_darwin64_record_arg_flush): Likewise.
9755 * config/rs6000/rs6000.c
9756 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9757 (rs6000_darwin64_record_arg_flush): Likewise.
9758 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9759 (sparc_function_value_1): Likewise.
9760 * config/spu/spu.c (adjust_operand): Likewise.
9761 (spu_emit_branch_or_set): Likewise.
9762 (arith_immediate_p): Likewise.
9763 * emit-rtl.c (gen_lowpart_common): Likewise.
9764 * expr.c (expand_expr_real_1): Likewise.
9765 * function.c (assign_parm_setup_block): Likewise.
9766 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9767 * reload1.c (alter_reg): Likewise.
9768 * stor-layout.c (mode_for_vector): Likewise.
9769 (layout_type): Likewise.
9771 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9773 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
9774 (spu_convert_move): Likewise.
9775 * lower-subreg.c (resolve_simple_move): Likewise.
9777 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9780 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
9781 define_insn to a define_expand.
9782 (altivec_vsum2sws_direct): New define_insn.
9783 (altivec_vsumsws): Convert from a define_insn to a define_expand.
9785 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9787 * config/arm/arm.c (arm_option_params_internal): Improve setting of
9790 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
9795 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
9796 realign stack if stack alignment needed is less than incoming
9799 2017-09-05 Marek Polacek <polacek@redhat.com>
9802 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
9805 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9807 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
9809 2017-09-05 Richard Biener <rguenther@suse.de>
9811 PR tree-optimization/82084
9812 * fold-const.c (can_native_encode_string_p): Handle wide characters.
9814 2017-09-05 Richard Biener <rguenther@suse.de>
9816 PR tree-optimization/82102
9817 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9819 2017-09-05 Martin Liska <mliska@suse.cz>
9821 PR tree-optimization/82032
9822 * tree-cfg.c (generate_range_test): New function.
9823 * tree-cfg.h (generate_range_test): Declared here.
9824 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9825 (cleanup_control_expr_graph): Use it.
9826 * tree-switch-conversion.c (try_switch_expansion): Remove
9828 (emit_case_nodes): Use generate_range_test.
9830 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9833 * config/i386/i386.md (*<btsc><mode>_mask): Add
9834 TARGET_USE_BT to insn constraint.
9835 (*btr<mode>_mask): Ditto.
9837 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9839 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9840 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9842 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9845 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9846 TARGET_NEON and TARGET_IWMMXT.
9847 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9848 TARGET_NEON and TARGET_IWMMXT.
9849 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9851 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9853 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9854 (ix86_rewrite_tls_address): Ditto.
9855 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9856 (ix86_rewrite_tls_address_1): Ditto.
9857 (ix86_rewrite_tls_address): Ditto.
9858 * config/i386/predicates.md (tls_address_pattern): New predicate.
9859 * config/i386/i386.md (TLS address splitter): New splitter.
9861 2017-09-04 Richard Biener <rguenther@suse.de>
9863 PR tree-optimization/82084
9864 * fold-const.h (can_native_encode_string_p): Declare.
9865 * fold-const.c (can_native_encode_string_p): Factor out from ...
9866 (native_encode_string): ... here.
9867 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9868 vectorizing stores from constants we later cannot handle.
9870 2017-09-04 Marek Polacek <polacek@redhat.com>
9873 * doc/invoke.texi: Update -Wtautological-compare documentation.
9875 2017-09-04 Jeff Law <law@redhat.com>
9877 PR tree-optimization/64910
9878 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9879 swap the first and last operand if the last is a constant.
9881 2017-09-04 Marek Polacek <polacek@redhat.com>
9884 * convert.c (do_narrow): When sanitizing signed integer overflows,
9885 bail out for signed types.
9886 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9888 2017-09-04 Richard Biener <rguenther@suse.de>
9890 PR tree-optimization/82060
9891 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9892 Move devirtualization after stmt folding and before EH/AB/noreturn
9893 cleanup to get the stmt refs canonicalized. Use a bool instead
9894 of gimple_modified_p since that doesn't work for NOPs. Schedule
9895 NOPs generated by folding for removal.
9897 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9898 Alan Hayward <alan.hayward@arm.com>
9899 David Sherwood <david.sherwood@arm.com>
9901 * coretypes.h (pad_direction): New enum.
9902 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9903 (FUNCTION_ARG_PADDING): Likewise.
9904 * target.def (function_arg_padding): New hook.
9905 * targhooks.h (default_function_arg_padding): Declare.
9906 * targhooks.c (default_function_arg_padding): New function.
9907 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9908 (TARGET_FUNCTION_ARG_PADDING): ...this.
9909 * doc/tm.texi: Regenerate.
9910 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9911 instead of direction.
9912 (compute_argument_addresses): Likewise.
9913 (load_register_parameters): Likewise.
9914 (emit_library_call_value_1): Likewise.
9915 (store_one_arg): Use targetm.calls.function_arg_padding instead
9916 of FUNCTION_ARG_PADDING.
9917 (must_pass_in_stack_var_size_or_pad): Likewise.
9918 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9919 (emit_group_store): Likewise.
9920 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9921 instead of FUNCTION_ARG_PADDING.
9922 (emit_push_insn): Likewise, and propagate enum change throughout
9924 * function.h (direction): Delete.
9925 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9927 * function.c (assign_parm_find_stack_rtl): Likewise.
9928 (assign_parm_setup_block_p): Likewise.
9929 (assign_parm_setup_block): Likewise.
9930 (gimplify_parameters): Likewise.
9931 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9932 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9934 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9935 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9936 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9937 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9938 (aarch64_function_arg_padding): ...this new function.
9939 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9940 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9941 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9942 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9943 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9944 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9945 (arm_pad_arg_upward): Replace with...
9946 (arm_function_arg_padding): ...this new function.
9947 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9949 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9950 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9951 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9952 (ia64_hpux_function_arg_padding): Replace with...
9953 (ia64_function_arg_padding): ...this new function. Use pad_direction
9954 instead of direction. Check for TARGET_HPUX.
9955 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9956 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9957 (iq2000_function_arg_padding): New function.
9958 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9959 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9960 (mips_function_arg_padding): ...this new function.
9961 (mips_pad_reg_upward): Update accordingly.
9962 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9963 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9964 targetm.calls.function_arg_padding.
9965 (FUNCTION_ARG_PADDING): Delete.
9966 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9967 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9968 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9969 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9970 (nios2_block_reg_padding): Return pad_direction instead of direction.
9971 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9972 instead of direction.
9973 (nios2_function_arg_padding): Likewise. Make static.
9974 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9975 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9976 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9977 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9978 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9979 (pa_function_arg_padding): Make static. Return pad_direction instead
9981 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9982 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9983 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9984 instead of direction. Use targetm.calls.function_arg_padding.
9985 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9986 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9987 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9988 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9989 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9991 (function_arg_padding): Rename to...
9992 (rs6000_function_arg_padding): ...this. Make static. Return
9993 pad_direction instead of direction.
9994 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9995 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9996 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9997 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9998 instead of direction. Use targetm.calls.function_arg_padding.
9999 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10000 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10001 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10002 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10003 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10004 (function_arg_padding): Rename to...
10005 (rs6000_function_arg_padding): ...this. Make static. Return
10006 pad_direction instead of direction.
10007 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10008 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10009 * config/s390/s390.c (s390_function_arg_padding): New function.
10010 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10011 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10012 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10013 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10014 (function_arg_padding): Rename to...
10015 (sparc_function_arg_padding): ...this. Make static. Return
10016 pad_direction instead of direction.
10017 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10018 * config/spu/spu.c (spu_function_arg_padding): New function.
10019 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10020 * system.h (FUNCTION_ARG_PADDING): Poison.
10022 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10023 Alan Hayward <alan.hayward@arm.com>
10024 David Sherwood <david.sherwood@arm.com>
10026 * target.def (modes_tieable_p): New hook.
10027 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10028 (TARGET_MODES_TIEABLE_P): ...this.
10029 * doc/tm.texi.in: Regenerate.
10030 * hooks.h (hook_bool_mode_mode_true): Declare.
10031 * hooks.c (hook_bool_mode_mode_true): New function.
10032 * combine.c (subst): Use targetm.modes_tieable_p instead of
10034 * dse.c (find_shift_sequence): Likewise.
10035 * expmed.c (extract_low_bits): Likewise.
10036 * lower-subreg.c: Include target.h.
10037 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10039 * rtlanal.c (rtx_cost): Likewise.
10040 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10041 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10042 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10043 (TARGET_MODES_TIEABLE_P): Redefine.
10044 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10045 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10046 (TARGET_MODES_TIEABLE_P): Redefine.
10047 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10048 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10049 (arc_modes_tieable_p): New function.
10050 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10051 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10052 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10053 (arm_modes_tieable_p): Make static.
10054 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10055 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10056 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10057 (TARGET_MODES_TIEABLE_P): Redefine.
10058 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10059 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10060 (TARGET_MODES_TIEABLE_P): Redefine.
10061 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10062 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10063 (cr16_modes_tieable_p): New function.
10064 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10065 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10066 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10067 (TRULY_NOOP_TRUNCATION): Update comment.
10068 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10069 (TRULY_NOOP_TRUNCATION): Update comment.
10070 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10071 (frv_modes_tieable_p): New function.
10072 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10073 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10074 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10075 (TARGET_MODES_TIEABLE_P): Redefine.
10076 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10077 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10078 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10079 (TARGET_MODES_TIEABLE_P): Redefine.
10080 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10081 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10082 (ia64_modes_tieable_p): New function.
10083 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10084 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10085 (iq2000_modes_tieable_p): New function.
10086 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10087 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10088 (lm32_modes_tieable_p): New function.
10089 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10090 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10091 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10092 (TARGET_MODES_TIEABLE_P): Redefine.
10093 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10094 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10095 (m32r_modes_tieable_p): New function.
10096 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10097 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10098 (m68k_modes_tieable_p): New function.
10099 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10100 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10101 (mcore_modes_tieable_p): New function.
10102 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10103 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10105 (TARGET_MODES_TIEABLE_P): Redefine.
10106 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10107 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10108 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10109 (TARGET_MODES_TIEABLE_P): Redefine.
10110 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10111 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10112 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10113 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10114 (mn10300_modes_tieable_p): ...this and make static.
10115 (TARGET_MODES_TIEABLE_P): Redefine.
10116 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10117 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10118 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10119 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10120 (msp430_modes_tieable_p): Make static.
10121 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10122 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10123 (TARGET_MODES_TIEABLE_P): Redefine.
10124 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10125 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10126 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10127 (TARGET_MODES_TIEABLE_P): Redefine.
10128 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10129 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10130 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10131 (TARGET_MODES_TIEABLE_P): Redefine.
10132 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10133 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10134 (pdp11_modes_tieable_p): New function.
10135 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10136 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10137 (rs6000_modes_tieable_p): New function.
10138 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10139 * config/powerpcspe/powerpcspe.md: Update comment.
10140 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10141 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10142 (TARGET_MODES_TIEABLE_P): Redefine.
10143 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10144 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10145 (rl78_modes_tieable_p): New function.
10146 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10147 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10148 (rs6000_modes_tieable_p): New function.
10149 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10150 * config/rs6000/rs6000.md: Update comment.
10151 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10152 * config/rx/rx.c (rx_modes_tieable_p): New function.
10153 (TARGET_MODES_TIEABLE_P): Redefine.
10154 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10155 * config/s390/s390.c (s390_modes_tieable_p): New function.
10156 (TARGET_MODES_TIEABLE_P): Redefine.
10157 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10158 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10159 (sh_modes_tieable_p): New function.
10160 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10161 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10162 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10163 (sparc_modes_tieable_p): Make static.
10164 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10165 * config/spu/spu.c (spu_modes_tieable_p): New function.
10166 (TARGET_MODES_TIEABLE_P): Redefine.
10167 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10168 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10169 (TARGET_MODES_TIEABLE_P): Redefine.
10170 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10171 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10172 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10173 * config/v850/v850.c (v850_modes_tieable_p): New function.
10174 (TARGET_MODES_TIEABLE_P): Redefine.
10175 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10176 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10177 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10178 (visium_modes_tieable_p): New function.
10179 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10180 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10181 (xtensa_modes_tieable_p): New function.
10182 * system.h (MODES_TIEABLE_P): Poison.
10184 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10185 Alan Hayward <alan.hayward@arm.com>
10186 David Sherwood <david.sherwood@arm.com>
10188 * target.def (hard_regno_mode_ok): New hook.
10189 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10190 (TARGET_HARD_REGNO_MODE_OK): ...this.
10191 * doc/tm.texi.in: Regenerate.
10192 * hooks.h (hook_bool_uint_mode_true): Declare.
10193 * hooks.c (hook_bool_uint_mode_true): New function.
10194 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10195 HARD_REGNO_MODE_OK.
10196 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10197 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10198 instead of HARD_REGNO_MODE_OK.
10199 * caller-save.c: Include target.h.
10200 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10201 HARD_REGNO_MODE_OK.
10202 * combine.c (can_combine_p): Likewise.
10203 (combinable_i3pat): Likewise.
10204 (can_change_dest_mode): Likewise.
10205 * expr.c (init_expr_target): Likewise.
10206 (convert_move): Likewise.
10207 (convert_modes): Likewise.
10208 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10209 (setup_prohibited_mode_move_regs): Likewise.
10210 * ira.h (target_ira): Likewise.
10211 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10212 * lra-constraints.c (process_alt_operands): Likewise.
10213 (split_reg): Likewise.
10214 * recog.c (peep2_find_free_register): Likewise.
10215 * ree.c (combine_reaching_defs): Likewise.
10216 * regcprop.c (maybe_mode_change): Likewise.
10217 * reginfo.c (init_reg_sets_1): Likewise.
10218 (choose_hard_reg_mode): Likewise.
10219 (simplifiable_subregs): Likewise.
10220 * regrename.c (check_new_reg_p): Likewise.
10221 * reload.c (find_valid_class): Likewise.
10222 (find_valid_class_1): Likewise.
10223 (reload_inner_reg_of_subreg): Likewise.
10224 (push_reload): Likewise.
10225 (combine_reloads): Likewise.
10226 (find_dummy_reload): Likewise.
10227 (find_reloads): Likewise.
10228 * reload1.c (find_reg): Likewise.
10229 (set_reload_reg): Likewise.
10230 (allocate_reload_reg): Likewise.
10231 (choose_reload_regs): Likewise.
10232 (reload_adjust_reg_for_temp): Likewise.
10233 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10234 (simplify_subreg_regno): Likewise.
10235 * sel-sched.c (init_regs_for_mode): Likewise.
10236 * varasm.c (make_decl_rtl): Likewise.
10237 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10238 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10239 HARD_REGNO_MODE_OK.
10240 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10241 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10242 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10243 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10244 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10245 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10246 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10247 (arc_mode_class): Delete.
10248 (HARD_REGNO_MODE_OK): Delete.
10249 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10250 (arc_hard_regno_mode_ok): Rename old array to...
10251 (arc_hard_regno_mode_ok_modes): ...this.
10252 (arc_conditional_register_usage): Update accordingly.
10253 (arc_mode_class): Make static.
10254 (arc_hard_regno_mode_ok): New function.
10255 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10256 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10257 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10258 (arm_hard_regno_mode_ok): Make static.
10259 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10260 HARD_REGNO_MODE_OK.
10261 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10262 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10263 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10265 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10266 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10267 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10268 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10269 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10270 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10271 * config/bfin/predicates.md (valid_reg_operand): Use
10272 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10273 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10274 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10275 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10276 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10277 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10278 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10279 (cr16_hard_regno_mode_ok): Make static and return a bool.
10280 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10281 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10282 (cris_hard_regno_mode_ok): New function.
10283 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10284 (epiphany_mode_class): Delete.
10285 (HARD_REGNO_MODE_OK): Delete.
10286 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10287 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10288 (hard_regno_mode_ok): Rename to...
10289 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10290 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10291 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10292 HARD_REGNO_MODE_OK.
10293 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10294 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10295 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10296 (frv_hard_regno_mode_ok): Make static and return a bool.
10297 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10298 HARD_REGNO_MODE_OK.
10299 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10300 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10301 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10302 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10304 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10305 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10306 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10307 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10309 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10310 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10311 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10312 (ia64_hard_regno_mode_ok): New function.
10313 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10314 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10315 (iq2000_hard_regno_mode_ok): New function.
10316 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10317 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10318 (lm32_hard_regno_mode_ok): New function.
10319 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10320 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10321 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10322 instead of HARD_REGNO_MODE_OK.
10323 (m32c_hard_regno_ok): Rename to...
10324 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10325 (m32c_cannot_change_mode_class): Update accordingly.
10326 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10327 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10328 (m32r_mode_class): Delete.
10329 (HARD_REGNO_MODE_OK): Delete.
10330 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10331 (m32r_hard_regno_mode_ok): Rename to...
10332 (m32r_hard_regno_modes): ...this.
10333 (m32r_mode_class): Make static.
10334 (m32r_hard_regno_mode_ok): New function.
10335 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10336 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10337 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10338 (m68k_hard_regno_mode_ok): Make static.
10339 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10340 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10341 (mcore_hard_regno_mode_ok): New function.
10342 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10343 (HARD_REGNO_MODE_OK): Delete.
10344 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10346 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10347 (microblaze_hard_regno_mode_ok): New function.
10348 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10349 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10350 (mips_hard_regno_mode_ok): Delete.
10351 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10352 (mips_hard_regno_mode_ok_p): ...this and make static.
10353 (mips_hard_regno_mode_ok_p): Rename to...
10354 (mips_hard_regno_mode_ok_uncached): ...this.
10355 (mips_hard_regno_mode_ok): New function.
10356 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10357 of HARD_REGNO_MODE_OK.
10358 (mips_option_override): Update after above name changes.
10359 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10360 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10361 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10362 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10363 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10364 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10365 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10366 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10367 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10368 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10369 (msp430_hard_regno_mode_ok): Make static and return a bool.
10370 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10371 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10372 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10374 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10375 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10376 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10377 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10378 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10379 (PA_HARD_REGNO_MODE_OK): ...this
10380 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10381 (PA_HARD_REGNO_MODE_OK): ...this.
10382 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10383 (pa_hard_regno_mode_ok): New function.
10384 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10385 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10386 (pdp11_hard_regno_mode_ok): New function.
10387 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10388 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10390 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10392 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10393 (rs6000_hard_regno_mode_ok): Rename to...
10394 (rs6000_hard_regno_mode_ok_uncached): ...this.
10395 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10396 (rs6000_hard_regno_mode_ok): New function.
10397 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10398 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10399 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10400 (riscv_hard_regno_mode_ok): ...this and make static.
10401 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10402 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10403 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10404 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10405 (rl78_hard_regno_mode_ok): Make static and return bool.
10406 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10407 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10409 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10410 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10411 (rs6000_hard_regno_mode_ok): Rename to...
10412 (rs6000_hard_regno_mode_ok_uncached): ...this.
10413 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10414 (rs6000_hard_regno_mode_ok): New function.
10415 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10416 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10417 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10418 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10419 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10420 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10421 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10422 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10423 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10424 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10425 (sh_hard_regno_mode_ok): Make static.
10426 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10427 instead of HARD_REGNO_MODE_OK.
10428 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10429 (sparc_mode_class): Delete.
10430 (HARD_REGNO_MODE_OK): Delete.
10431 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10432 (hard_regno_mode_classes): Make static.
10433 (sparc_mode_class): Likewise.
10434 (sparc_hard_regno_mode_ok): New function.
10435 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10436 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10437 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10439 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10440 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10441 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10442 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10443 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10444 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10445 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10446 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10447 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10448 (visium_hard_regno_mode_ok): New function.
10449 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10450 instead of HARD_REGNO_MODE_OK.
10451 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10452 (HARD_REGNO_MODE_OK): Delete.
10453 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10454 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10455 (xtensa_option_override): Update accordingly.
10456 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10457 (xtensa_hard_regno_mode_ok): New function.
10458 * system.h (HARD_REGNO_MODE_OK): Poison.
10460 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10461 Alan Hayward <alan.hayward@arm.com>
10462 David Sherwood <david.sherwood@arm.com>
10464 * target.def (hard_regno_call_part_clobbered): New hook.
10465 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10466 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10467 * doc/tm.texi: Regenerate.
10468 * hooks.h (hook_bool_uint_mode_false): Declare.
10469 * hooks.c (hook_bool_uint_mode_false): New function.
10470 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10471 * cselib.c (cselib_process_insn): Use
10472 targetm.hard_regno_call_part_clobbered instead of
10473 HARD_REGNO_CALL_PART_CLOBBERED.
10474 * ira-conflicts.c (ira_build_conflicts): Likewise.
10475 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10476 * lra-constraints.c (need_for_call_save_p): Likewise.
10477 * lra-lives.c: Include target.h.
10478 (check_pseudos_live_through_calls): Use
10479 targetm.hard_regno_call_part_clobbered instead of
10480 HARD_REGNO_CALL_PART_CLOBBERED.
10481 * regcprop.c: Include target.h.
10482 (copyprop_hardreg_forward_1): Use
10483 targetm.hard_regno_call_part_clobbered instead of
10484 HARD_REGNO_CALL_PART_CLOBBERED.
10485 * reginfo.c (choose_hard_reg_mode): Likewise.
10486 * regrename.c (check_new_reg_p): Likewise.
10487 * reload.c (find_equiv_reg): Likewise.
10488 * reload1.c (emit_reload_insns): Likewise.
10489 * sched-deps.c (deps_analyze_insn): Likewise.
10490 * sel-sched.c (init_regs_for_mode): Likewise.
10491 (mark_unavailable_hard_regs): Likewise.
10492 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10493 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10494 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10496 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10497 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10498 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10500 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10502 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10503 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10504 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10506 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10507 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10508 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10510 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10511 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10513 * config/powerpcspe/powerpcspe.c
10514 (rs6000_hard_regno_call_part_clobbered): New function.
10515 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10516 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10517 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10519 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10520 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10521 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10523 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10524 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10525 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10527 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10528 Alan Hayward <alan.hayward@arm.com>
10529 David Sherwood <david.sherwood@arm.com>
10531 * rtl.h (subreg_memory_offset): Declare.
10532 * emit-rtl.c (subreg_memory_offset): New function.
10533 * expmed.c (store_bit_field_1): Use it.
10534 * expr.c (undefined_operand_subword_p): Likewise.
10535 * simplify-rtx.c (simplify_subreg): Likewise.
10537 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10539 PR rtl-optimization/57448
10542 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10543 using atomic_load pattern.
10544 (expand_atomic_store): Likewise.
10546 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10549 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10550 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10551 handling. Use replace_call_with_value with NULL instead of
10552 gsi_replace, unlink_stmt_vdef and release_defs.
10554 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
10557 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
10559 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10562 * rtl.h (emit_library_call_value_1): Declare.
10563 (emit_library_call): Replace declaration with a series of overloads.
10564 Remove the parameter count argument.
10565 (emit_library_call_value): Likewise.
10566 * calls.c (emit_library_call_value_1): Make global. Replace varargs
10567 with an "rtx_mode_t *".
10568 (emit_library_call_value): Delete.
10569 (emit_library_call): Likewise.
10570 * asan.c (asan_emit_stack_protection): Update calls accordingly.
10571 (asan_emit_allocas_unpoison): Likewise.
10572 * builtins.c (expand_builtin_powi): Likewise.
10573 (expand_asan_emit_allocas_unpoison): Likewise.
10574 * cfgexpand.c (expand_main_function): Likewise.
10575 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
10576 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
10577 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
10578 * config/arm/arm.c (arm_trampoline_init): Likewise.
10579 (arm_call_tls_get_addr): Likewise.
10580 (arm_expand_divmod_libfunc): Likewise.
10581 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
10582 (smulsi3_highpart): Likewise.
10583 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
10584 (c6x_expand_compare): Likewise.
10585 (c6x_expand_movmem): Likewise.
10586 * config/frv/frv.c (frv_trampoline_init): Likewise.
10587 * config/i386/i386.c (ix86_trampoline_init): Likewise.
10588 (ix86_expand_divmod_libfunc): Likewise.
10589 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
10590 (ia64_expand_compare): Likewise.
10591 (ia64_profile_hook): Likewise.
10592 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
10593 (nonlocal_goto): Likewise.
10594 (restore_stack_nonlocal): Likewise.
10595 * config/m32r/m32r.c (block_move_call): Likewise.
10596 (m32r_trampoline_init): Likewise.
10597 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
10598 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
10599 (m68k_call_m68k_read_tp): Likewise.
10600 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
10601 (microblaze_expand_divide): Likewise.
10602 * config/mips/mips.h (mips_args): Likewise.
10603 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
10604 (MIPS_ICACHE_SYNC): Likewise.
10605 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
10606 (nios2_trampoline_init): Likewise.
10607 * config/pa/pa.c (hppa_tls_call): Likewise.
10608 (pa_trampoline_init): Likewise.
10609 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
10610 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
10611 (expand_strn_compare): Likewise.
10612 (rs6000_generate_compare): Likewise.
10613 (rs6000_expand_float128_convert): Likewise.
10614 (output_profile_hook): Likewise.
10615 (rs6000_trampoline_init): Likewise.
10616 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
10617 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
10618 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
10619 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
10620 (rs6000_generate_compare): Likewise.
10621 (rs6000_expand_float128_convert): Likewise.
10622 (output_profile_hook): Likewise.
10623 (rs6000_trampoline_init): Likewise.
10624 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
10625 * config/sh/sh.c (sh_trampoline_init): Likewise.
10626 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
10627 (sparc_emit_float_lib_cmp): Likewise.
10628 (sparc32_initialize_trampoline): Likewise.
10629 (sparc64_initialize_trampoline): Likewise.
10630 (sparc_profile_hook): Likewise.
10631 * config/spu/spu.c (ea_load_store): Likewise.
10632 * config/spu/spu.md (floatunssidf2): Likewise.
10633 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
10634 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
10635 * config/visium/visium.c (expand_block_move_4): Likewise.
10636 (expand_block_move_2): Likewise.
10637 (expand_block_move_1): Likewise.
10638 (expand_block_set_4): Likewise.
10639 (expand_block_set_2): Likewise.
10640 (expand_block_set_1): Likewise.
10641 (visium_trampoline_init): Likewise.
10642 (visium_profile_hook): Likewise.
10643 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
10644 (xtensa_setup_frame_addresses): Likewise.
10645 (xtensa_trampoline_init): Likewise.
10646 * except.c (sjlj_emit_function_enter): Likewise.
10647 (sjlj_emit_function_exit): Likewise.
10648 * explow.c (allocate_dynamic_stack_space): Likewise.
10649 (probe_stack_range): Likewise.
10650 * expr.c (convert_mode_scalar): Likewise.
10651 * optabs.c (expand_binop): Likewise.
10652 (expand_twoval_binop_libfunc): Likewise.
10653 (expand_unop): Likewise.
10654 (prepare_cmp_insn): Likewise.
10655 (prepare_float_lib_cmp): Likewise.
10656 (expand_float): Likewise.
10657 (expand_fix): Likewise.
10658 (expand_fixed_convert): Likewise.
10659 (maybe_emit_sync_lock_test_and_set): Likewise.
10660 (expand_atomic_compare_and_swap): Likewise.
10661 (expand_mem_thread_fence): Likewise.
10662 (expand_atomic_fetch_op): Likewise.
10664 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
10666 * doc/generic.texi (OpenACC): Adjust URL.
10667 * doc/invoke.texi (C Dialect Options): Ditto.
10669 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
10671 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
10672 predicate for operand 1. Add (m,<S>) constraint.
10673 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
10674 Prevent memory operand 1 with register operand 2.
10676 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
10678 PR rtl-optimization/82024
10679 * combine.c (try_combine): If the combination result is a PARALLEL,
10680 and we only need to retain the SET in there that would be placed
10681 at I2, check that we can place that at I3 instead, before doing so.
10683 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10686 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
10688 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
10689 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
10692 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
10693 Jeff Law <law@redhat.com>
10695 * varasm.c (bss_initializer_p): Do not put constants into .bss
10696 (categorize_decl_for_section): Handle bss_initializer_p returning
10697 false when DECL_INITIAL is NULL.
10699 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10702 * config/s390/s390.c (s390_can_inline_p): New function.
10704 2017-09-01 Jeff Law <law@redhat.com>
10706 PR tree-optimization/82052
10707 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
10708 Always initialize the returned slot after a hash table miss
10709 when INSERT is true.
10711 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
10713 * config/s390/s390.md (mem_signal_fence): Remove.
10714 * doc/md.texi (mem_signal_fence): Remove.
10715 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
10717 * target-insns.def (mem_signal_fence): Remove.
10719 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10722 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
10725 * asan.c (create_odr_indicator): Strip name encoding from assembler
10726 name before appending it after __odr_asan_.
10728 2017-09-01 Martin Liska <mliska@suse.cz>
10730 PR tree-optimization/82059
10731 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
10732 frequency only when an edge is redirected.
10734 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10736 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
10737 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
10738 (arc_conditional_register_usage): Remove ARC600 lp_count
10740 (arc_file_start): Emit Tag_ARC_CPU_variation.
10741 (arc_can_use_doloop_p): New conditions to use ZOLs.
10742 (hwloop_fail): New function.
10743 (hwloop_optimize): Likewise.
10744 (hwloop_pattern_reg): Likewise.
10745 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10746 (arc_reorg_loops): New function, calls reorg_loops.
10747 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10748 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10750 (arc_loop_hazard): Remove function, functionality moved into
10752 (arc_hazard): Remove arc_loop_hazard call.
10753 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10754 into hwloop_optimize.
10755 (arc_label_align): Remove ZOL handling.
10756 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10757 * config/arc/arc.md (doloop_begin): Remove pattern.
10758 (doloop_begin_i): Likewise.
10759 (doloop_end_i): Likewise.
10760 (doloop_fallback): Likewise.
10761 (doloop_fallback_m): Likewise.
10762 (doloop_end): Reimplement expand.
10763 (arc_lp): New pattern for LP instruction.
10764 (loop_end): New pattern.
10765 (loop_fail): Likewise.
10766 (decrement_and_branch_until_zero): Likewise.
10767 * config/arc/arc.opt (mlpc-width): New option.
10768 * doc/invoke.texi (mlpc-width): Document option.
10770 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10772 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
10773 (arc_ccfsm_advance): Fix checking for delay slots.
10774 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
10776 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10778 * config/arc/arc.md (movqi_insn): Add stores to save constant long
10780 (movhi_insn): Update store instruction constraint which are saving
10781 6-bit short immediates.
10782 (movsi_insn): Consider also short scaled load operations.
10783 (zero_extendhisi2_i): Use Usd constraint instead of T.
10784 (extendhisi2_i): Add q constraint.
10785 (arc_clzsi2): Add type and length attributes.
10786 (arc_ctzsi2): Likewise.
10787 * config/arc/constraints.md (Usc): Update constraint, the
10788 assembler can parse two relocations for a single instruction.
10790 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10792 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
10793 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
10795 2017-08-31 Olivier Hainque <hainque@adacore.com>
10797 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
10798 match as powerpc-wrs-vxworks*.
10800 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
10802 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
10803 register constraint for by-element operand.
10804 (aarch64_mls_elt_merge<mode>): Likewise.
10806 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10808 * config/arc/arc.c (arc_can_follow_jump): Check for short
10811 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10813 * config.gcc: Use g.opt for arc.
10814 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
10815 functionality moved to ...
10816 (legitimate_scaled_address_p): New function, ...here.
10817 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
10818 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10819 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10821 (arc_override_options): Handle G option.
10822 (arc_output_pic_addr_const): Correct function definition.
10823 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10824 (arc_decl_anon_ns_mem_p): Delete.
10825 (arc_in_small_data_p): Overhaul this function to take into
10826 consideration the value given via G option.
10827 (arc_rewrite_small_data_1): Renamed and corrected old
10828 arc_rewrite_small_data function.
10829 (arc_rewrite_small_data): New function.
10830 (small_data_pattern): Don't use pic_offset_table_rtx.
10831 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10832 * config/arc/simdext.md (movmisalignv2hi): Use
10833 prepare_move_operands function.
10835 (movmisalign*): Likewise.
10836 * doc/invoke.texi (ARC options): Document -G option.
10838 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10840 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10842 * config/arc/arc.c (arc_print_operand): Output scalled address for
10843 sdata whenever is possible.
10844 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10845 load/stores are available.
10846 (compact_sda_memory_operand): Check for the alignment required by
10847 code density instructions.
10848 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10850 * config/arc/constraints.md (Usd): Update constraint.
10851 (Us0): New constraint.
10852 (Usc): Update constraint.
10854 2017-08-31 Richard Biener <rguenther@suse.de>
10856 PR middle-end/82054
10857 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10858 function only once.
10860 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10862 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10863 Resize type_signature.
10865 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10866 Alan Hayward <alan.hayward@arm.com>
10867 David Sherwood <david.sherwood@arm.com>
10869 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10870 subregs whose inner modes can be stored in GPRs.
10871 (aarch64_classify_index): Likewise.
10873 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10874 Alan Hayward <alan.hayward@arm.com>
10875 David Sherwood <david.sherwood@arm.com>
10877 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10878 (V_INT_EQUIV): ...this.
10879 (v_cmp_result): Rename to...
10880 (v_int_equiv): ...this.
10881 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10882 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10883 (copysign<mode>3): Likewise.
10884 (aarch64_simd_bsl<mode>_internal): Likewise.
10885 (aarch64_simd_bsl<mode>): Likewise.
10886 (vec_cmp<mode><mode>): Likewise.
10887 (vcond<mode><mode>): Likewise.
10888 (vcond<v_cmp_mixed><mode>): Likewise.
10889 (vcondu<mode><v_cmp_mixed>): Likewise.
10890 (aarch64_cm<optab><mode>): Likewise.
10891 (aarch64_cmtst<mode>): Likewise.
10892 (aarch64_fac<optab><mode>): Likewise.
10893 (vec_perm_const<mode>): Likewise.
10894 (vcond_mask_<mode><v_cmp_result>): Rename to...
10895 (vcond_mask_<mode><v_int_equiv>): ...this.
10896 (vec_cmp<mode><v_cmp_result>): Rename to...
10897 (vec_cmp<mode><v_int_equiv>): ...this.
10899 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10900 Alan Hayward <alan.hayward@arm.com>
10901 David Sherwood <david.sherwood@arm.com>
10903 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10905 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10906 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10907 (UNSPEC_LD4_DREG): New unspecs.
10908 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10909 (aarch64_ld2<mode>_dreg_be): Replace with...
10910 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10912 (aarch64_ld3<mode>_dreg_le)
10913 (aarch64_ld3<mode>_dreg_be): Replace with...
10914 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10916 (aarch64_ld4<mode>_dreg_le)
10917 (aarch64_ld4<mode>_dreg_be): Replace with...
10918 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10921 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10923 PR tree-optimization/81987
10924 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10925 insert an initializer in a location not dominated by the stride
10928 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10930 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10931 on the entire header of the finally block in the fallthru case.
10933 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10935 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10937 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10939 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10940 rs6000_emit_move_from_cr and call renamed function.
10941 (rs6000_emit_prologue): Call renamed functions.
10942 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10943 movesi_from_cr, remove volatile CRs.
10945 2017-08-30 Jon Beniston <jon@beniston.com>
10946 Richard Biener <rguenther@suse.de>
10948 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10949 of VECTOR_MODE_P check.
10950 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10951 element vector types.
10953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10955 * df.h (df_read_modify_subreg_p): Remove in favor of...
10956 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10957 const_rtx instead of an rtx.
10958 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10959 * df-problems.c (df_word_lr_mark_ref): Likewise.
10960 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10961 (mark_pseudo_reg_dead): Likewise.
10962 (mark_ref_dead): Likewise.
10963 * reginfo.c (init_subregs_of_mode): Likewise.
10964 * sched-deps.c (sched_analyze_1): Likewise.
10965 * df-scan.c (df_def_record_1): Likewise.
10966 (df_uses_record): Likewise.
10967 (df_read_modify_subreg_p): Remove in favor of...
10968 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10969 const_rtx instead of an rtx.
10971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10972 Alan Hayward <alan.hayward@arm.com>
10973 David Sherwood <david.sherwood@arm.com>
10975 * rtl.h (partial_subreg_p): New function.
10976 * caller-save.c (save_call_clobbered_regs): Use it.
10977 * calls.c (expand_call): Likewise.
10978 * combine.c (combinable_i3pat): Likewise.
10979 (simplify_set): Likewise.
10980 (make_extraction): Likewise.
10981 (make_compound_operation_int): Likewise.
10982 (gen_lowpart_or_truncate): Likewise.
10983 (force_to_mode): Likewise.
10984 (make_field_assignment): Likewise.
10985 (reg_truncated_to_mode): Likewise.
10986 (record_truncated_value): Likewise.
10987 (move_deaths): Likewise.
10988 * cse.c (record_jump_cond): Likewise.
10989 (cse_insn): Likewise.
10990 * cselib.c (cselib_lookup_1): Likewise.
10991 * expmed.c (extract_bit_field_using_extv): Likewise.
10992 * function.c (assign_parm_setup_reg): Likewise.
10993 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10994 * ira-build.c (create_insn_allocnos): Likewise.
10995 * lra-coalesce.c (merge_pseudos): Likewise.
10996 * lra-constraints.c (match_reload): Likewise.
10997 (simplify_operand_subreg): Likewise.
10998 (curr_insn_transform): Likewise.
10999 * lra-lives.c (process_bb_lives): Likewise.
11000 * lra.c (new_insn_reg): Likewise.
11001 (lra_substitute_pseudo): Likewise.
11002 * regcprop.c (mode_change_ok): Likewise.
11003 (maybe_mode_change): Likewise.
11004 (copyprop_hardreg_forward_1): Likewise.
11005 * reload.c (push_reload): Likewise.
11006 (find_reloads): Likewise.
11007 (find_reloads_subreg_address): Likewise.
11008 * reload1.c (alter_reg): Likewise.
11009 (eliminate_regs_1): Likewise.
11010 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11012 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11014 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11015 back to if statements, including unpack.
11017 2017-08-30 Martin Liska <mliska@suse.cz>
11019 PR inline-asm/82001
11020 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11022 (func_checker::compare_asm_inputs_outputs): ... this function.
11023 (func_checker::compare_gimple_asm): Use the function to compare
11024 also ASM constrains.
11025 * ipa-icf-gimple.h: Rename the function.
11027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11028 Alan Hayward <alan.hayward@arm.com>
11029 David Sherwood <david.sherwood@arm.com>
11031 * coretypes.h (complex_mode): New type.
11032 * gdbhooks.py (build_pretty_printer): Handle it.
11033 * machmode.h (complex_mode): New class.
11034 (complex_mode::includes_p): New function.
11035 (is_complex_int_mode): Likewise.
11036 (is_complex_float_mode): Likewise.
11037 * genmodes.c (get_mode_class): Handle complex mode classes.
11038 * function.c (expand_function_end): Use is_complex_int_mode.
11040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11041 Alan Hayward <alan.hayward@arm.com>
11042 David Sherwood <david.sherwood@arm.com>
11044 * coretypes.h (scalar_mode_pod): New typedef.
11045 * gdbhooks.py (build_pretty_printer): Handle it.
11046 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11047 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11048 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11049 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11050 as_a <scalar_mode>.
11052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11053 Alan Hayward <alan.hayward@arm.com>
11054 David Sherwood <david.sherwood@arm.com>
11056 * machmode.h (mode_for_vector): Take a scalar_mode instead
11058 * stor-layout.c (mode_for_vector): Likewise.
11059 * explow.c (promote_mode): Use as_a <scalar_mode>.
11060 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11063 Alan Hayward <alan.hayward@arm.com>
11064 David Sherwood <david.sherwood@arm.com>
11066 * target.def (preferred_simd_mode): Take a scalar_mode
11067 instead of a machine_mode.
11068 * targhooks.h (default_preferred_simd_mode): Likewise.
11069 * targhooks.c (default_preferred_simd_mode): Likewise.
11070 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11071 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11072 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11073 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11074 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11075 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11076 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11077 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11079 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11080 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11081 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11082 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11083 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11084 * doc/tm.texi: Regenerate.
11085 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11089 Alan Hayward <alan.hayward@arm.com>
11090 David Sherwood <david.sherwood@arm.com>
11092 * target.def (scalar_mode_supported_p): Take a scalar_mode
11093 instead of a machine_mode.
11094 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11095 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11096 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11097 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11098 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11099 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11100 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11101 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11102 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11103 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11104 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11105 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11106 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11107 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11109 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11110 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11111 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11112 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11113 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11115 * doc/tm.texi: Regenerate.
11117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11118 Alan Hayward <alan.hayward@arm.com>
11119 David Sherwood <david.sherwood@arm.com>
11121 * coretypes.h (opt_scalar_mode): New typedef.
11122 * gdbhooks.py (build_pretty_printers): Handle it.
11123 * machmode.h (mode_iterator::get_2xwider): Add overload for
11125 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11127 * expr.c (convert_mode_scalar): Likewise.
11128 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11129 * optabs.c (expand_float): Likewise.
11130 (expand_fix): Likewise.
11131 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11134 Alan Hayward <alan.hayward@arm.com>
11135 David Sherwood <david.sherwood@arm.com>
11137 * optabs.c (expand_float): Explicitly check for scalars before
11138 using a branching expansion.
11139 (expand_fix): Likewise.
11141 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11142 Alan Hayward <alan.hayward@arm.com>
11143 David Sherwood <david.sherwood@arm.com>
11145 * expr.c (convert_mode): Split scalar handling out into...
11146 (convert_mode_scalar): ...this new function. Treat the modes
11149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11150 Alan Hayward <alan.hayward@arm.com>
11151 David Sherwood <david.sherwood@arm.com>
11153 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11155 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11158 Alan Hayward <alan.hayward@arm.com>
11159 David Sherwood <david.sherwood@arm.com>
11161 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11162 rather than a machine_mode.
11163 (fixed_from_string): Likewise.
11164 (fixed_convert): Likewise.
11165 (fixed_convert_from_int): Likewise.
11166 (fixed_convert_from_real): Likewise.
11167 (real_convert_from_fixed): Likewise.
11168 * fixed-value.c (fixed_from_double_int): Likewise.
11169 (fixed_from_string): Likewise.
11170 (fixed_convert): Likewise.
11171 (fixed_convert_from_int): Likewise.
11172 (fixed_convert_from_real): Likewise.
11173 (real_convert_from_fixed): Likewise.
11174 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11177 Alan Hayward <alan.hayward@arm.com>
11178 David Sherwood <david.sherwood@arm.com>
11180 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11181 of separate mode class checks. Do not allow vector modes here.
11182 (immed_wide_int_const): Use as_a <scalar_mode>.
11183 * explow.c (trunc_int_for_mode): Likewise.
11184 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11185 (wi::shwi): Likewise.
11186 (wi::min_value): Likewise.
11187 (wi::max_value): Likewise.
11188 * dwarf2out.c (loc_descriptor): Likewise.
11189 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11190 for CONST_WIDE_INT.
11192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11193 Alan Hayward <alan.hayward@arm.com>
11194 David Sherwood <david.sherwood@arm.com>
11196 * tree.h (SCALAR_TYPE_MODE): New macro.
11197 * expr.c (expand_expr_addr_expr_1): Use it.
11198 (expand_expr_real_2): Likewise.
11199 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11200 (fold_convert_const_fixed_from_int): Likewise.
11201 (fold_convert_const_fixed_from_real): Likewise.
11202 (native_encode_fixed): Likewise
11203 (native_encode_complex): Likewise
11204 (native_encode_vector): Likewise.
11205 (native_interpret_fixed): Likewise.
11206 (native_interpret_real): Likewise.
11207 (native_interpret_complex): Likewise.
11208 (native_interpret_vector): Likewise.
11209 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11210 (simd_clone_adjust_argument_types): Likewise.
11211 (simd_clone_init_simd_arrays): Likewise.
11212 (simd_clone_adjust): Likewise.
11213 * stor-layout.c (layout_type): Likewise.
11214 * tree.c (build_minus_one_cst): Likewise.
11215 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11216 * tree-inline.c (estimate_move_cost): Likewise.
11217 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11218 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11219 (vectorizable_reduction): Likewise.
11220 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11221 (vect_recog_mixed_size_cond_pattern): Likewise.
11222 (check_bool_pattern): Likewise.
11223 (adjust_bool_pattern): Likewise.
11224 (search_type_for_mask_1): Likewise.
11225 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11226 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11227 (vectorizable_load): Likewise.
11228 (vectorizable_store): Likewise.
11229 * ubsan.c (ubsan_encode_value): Likewise.
11230 * varasm.c (output_constant): Likewise.
11232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11233 Alan Hayward <alan.hayward@arm.com>
11234 David Sherwood <david.sherwood@arm.com>
11236 * coretypes.h (scalar_mode): New class.
11237 * machmode.h (scalar_mode): Likewise.
11238 (scalar_mode::includes_p): New function.
11239 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11240 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11241 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11242 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11243 * expmed.c (store_bit_field_1): Likewise.
11244 (extract_bit_field_1): Likewise.
11245 * expr.c (write_complex_part): Likewise.
11246 (read_complex_part): Likewise.
11247 (emit_move_complex_push): Likewise.
11248 (expand_expr_real_2): Likewise.
11249 * function.c (assign_parm_setup_reg): Likewise.
11250 (assign_parms_unsplit_complex): Likewise.
11251 * optabs.c (expand_binop): Likewise.
11252 * rtlanal.c (subreg_get_info): Likewise.
11253 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11254 * varasm.c (output_constant_pool_2): Likewise.
11256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11257 Alan Hayward <alan.hayward@arm.com>
11258 David Sherwood <david.sherwood@arm.com>
11260 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11262 (expmed_mult_highpart_optab): Likewise.
11263 (expmed_mult_highpart): Likewise.
11265 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11266 Alan Hayward <alan.hayward@arm.com>
11267 David Sherwood <david.sherwood@arm.com>
11269 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11270 instead of a machine_mode.
11271 (builtin_memset_read_str): Likewise.
11272 * builtins.c (c_readstr): Likewise.
11273 (builtin_memcpy_read_str): Likewise.
11274 (builtin_strncpy_read_str): Likewise.
11275 (builtin_memset_read_str): Likewise.
11276 (builtin_memset_gen_str): Likewise.
11277 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11278 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11279 instead of a machine_mode.
11280 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11282 (make_extraction): Likewise.
11283 (try_widen_shift_mode): Take and return scalar_int_modes instead
11285 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11286 a scalar_int_mode instead of a machine_mode.
11287 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11288 (avr_addr_space_pointer_mode): Likewise.
11289 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11290 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11291 (msp430_unwind_word_mode): Likewise.
11292 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11293 (spu_addr_space_pointer_mode): Likewise.
11294 (spu_addr_space_address_mode): Likewise.
11295 (spu_libgcc_cmp_return_mode): Likewise.
11296 (spu_libgcc_shift_count_mode): Likewise.
11297 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11298 (rl78_addr_space_pointer_mode): Likewise.
11299 (fl78_unwind_word_mode): Likewise.
11300 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11302 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11303 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11304 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11305 (mips_valid_pointer_mode): Likewise.
11306 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11307 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11308 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11310 (ft32_addr_space_address_mode): Likewise.
11311 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11312 scalar_int_mode instead of a machine_mode.
11313 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11315 (m32c_addr_space_address_mode): Likewise.
11316 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11317 (rs6000_eh_return_filter_mode): Likewise.
11318 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11319 (rs6000_eh_return_filter_mode): Likewise.
11320 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11321 (s390_libgcc_shift_count_mode): Likewise.
11322 (s390_unwind_word_mode): Likewise.
11323 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11325 * target.def (mode_rep_extended): Likewise.
11326 (valid_pointer_mode): Likewise.
11327 (addr_space.valid_pointer_mode): Likewise.
11328 (eh_return_filter_mode): Return a scalar_int_mode rather than
11330 (libgcc_cmp_return_mode): Likewise.
11331 (libgcc_shift_count_mode): Likewise.
11332 (unwind_word_mode): Likewise.
11333 (addr_space.pointer_mode): Likewise.
11334 (addr_space.address_mode): Likewise.
11335 * doc/tm.texi: Regenerate.
11336 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11338 (do_jump): Use scalar_int_mode for local variables.
11339 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11340 rather than a machine_mode.
11341 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11342 (scompare_loc_descriptor_wide): Likewise.
11343 (scompare_loc_descriptor_narrow): Likewise.
11344 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11346 * except.c (sjlj_emit_dispatch_table): Likewise.
11347 (expand_builtin_eh_copy_values): Likewise.
11348 * explow.c (convert_memory_address_addr_space_1): Likewise.
11349 Take a scalar_int_mode rather than a machine_mode.
11350 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11351 than a machine_mode.
11352 (memory_address_addr_space): Use scalar_int_mode for local variables.
11353 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11354 rather than a machine_mode.
11355 * expmed.c (mask_rtx): Likewise.
11356 (init_expmed_one_conv): Likewise.
11357 (expand_mult_highpart_adjust): Likewise.
11358 (extract_high_half): Likewise.
11359 (expmed_mult_highpart_optab): Likewise.
11360 (expmed_mult_highpart): Likewise.
11361 (expand_smod_pow2): Likewise.
11362 (expand_sdiv_pow2): Likewise.
11363 (emit_store_flag_int): Likewise.
11364 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11366 (extract_low_bits): Likewise.
11367 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11369 * expr.c (pieces_addr::adjust): Likewise.
11370 (can_store_by_pieces): Likewise.
11371 (store_by_pieces): Likewise.
11372 (clear_by_pieces_1): Likewise.
11373 (expand_expr_addr_expr_1): Likewise.
11374 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11375 (expand_expr_real_1): Likewise.
11376 (try_casesi): Likewise.
11377 * final.c (shorten_branches): Likewise.
11378 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11379 type of "mode" to machine_mode.
11380 * internal-fn.c (expand_arith_overflow_result_store): Take a
11381 scalar_int_mode rather than a machine_mode.
11382 (expand_mul_overflow): Use scalar_int_mode for local variables.
11383 * loop-doloop.c (doloop_modify): Likewise.
11384 (doloop_optimize): Likewise.
11385 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11386 than a machine_mode.
11387 (expand_doubleword_shift_condmove): Likewise.
11388 (expand_doubleword_shift): Likewise.
11389 (expand_doubleword_clz): Likewise.
11390 (expand_doubleword_popcount): Likewise.
11391 (expand_doubleword_parity): Likewise.
11392 (expand_absneg_bit): Use scalar_int_mode for local variables.
11393 (prepare_float_lib_cmp): Likewise.
11394 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11395 rather than a machine_mode.
11396 (convert_memory_address_addr_space): Likewise.
11397 (get_mode_bounds): Likewise.
11398 (get_address_mode): Return a scalar_int_mode rather than a
11400 * rtlanal.c (get_address_mode): Likewise.
11401 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11402 than a machine_mode.
11403 * targhooks.c (default_mode_rep_extended): Likewise.
11404 (default_valid_pointer_mode): Likewise.
11405 (default_addr_space_valid_pointer_mode): Likewise.
11406 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11407 than a machine_mode.
11408 (default_libgcc_cmp_return_mode): Likewise.
11409 (default_libgcc_shift_count_mode): Likewise.
11410 (default_unwind_word_mode): Likewise.
11411 (default_addr_space_pointer_mode): Likewise.
11412 (default_addr_space_address_mode): Likewise.
11413 * targhooks.h (default_eh_return_filter_mode): Likewise.
11414 (default_libgcc_cmp_return_mode): Likewise.
11415 (default_libgcc_shift_count_mode): Likewise.
11416 (default_unwind_word_mode): Likewise.
11417 (default_addr_space_pointer_mode): Likewise.
11418 (default_addr_space_address_mode): Likewise.
11419 (default_mode_rep_extended): Take a scalar_int_mode rather than
11421 (default_valid_pointer_mode): Likewise.
11422 (default_addr_space_valid_pointer_mode): Likewise.
11423 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11425 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11426 rather than a machine_mode.
11427 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11428 for local variables.
11429 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11430 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11431 than a machine_mode.
11433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11434 Alan Hayward <alan.hayward@arm.com>
11435 David Sherwood <david.sherwood@arm.com>
11437 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11438 the mode argument to scalar_int_mode.
11439 (do_jump_by_parts_zero_rtx): Likewise.
11440 (do_jump_by_parts_equality_rtx): Likewise.
11441 (do_jump_by_parts_greater): Take a mode argument.
11442 (do_jump_by_parts_equality): Likewise.
11443 (do_jump_1): Update calls accordingly.
11445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11446 Alan Hayward <alan.hayward@arm.com>
11447 David Sherwood <david.sherwood@arm.com>
11449 * is-a.h (safe_dyn_cast): New function.
11450 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11451 (jump_table_for_label): Likewise.
11452 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11453 instead of an rtx_insn *.
11454 (shorten_branches): Use dyn_cast instead of LABEL_P and
11455 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11456 rtx_jump_table_data::get_data_mode.
11457 (final_scan_insn): Likewise.
11459 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11460 Alan Hayward <alan.hayward@arm.com>
11461 David Sherwood <david.sherwood@arm.com>
11463 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11464 trying to combine a full-register integer set with a subreg
11467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11468 Alan Hayward <alan.hayward@arm.com>
11469 David Sherwood <david.sherwood@arm.com>
11471 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11472 that is always either address_mode or pointer_mode.
11474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11475 Alan Hayward <alan.hayward@arm.com>
11476 David Sherwood <david.sherwood@arm.com>
11478 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11479 when the two are known to be equal.
11481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11482 Alan Hayward <alan.hayward@arm.com>
11483 David Sherwood <david.sherwood@arm.com>
11485 * simplify-rtx.c (simplify_const_unary_operation): Use
11486 is_a <scalar_int_mode> instead of checking for a nonzero
11487 precision. Forcibly convert op_mode to a scalar_int_mode
11488 in that case. More clearly differentiate the operand and
11489 result modes and use the former when deciding what the value
11490 of a count-bits operation should be. Use is_int_mode instead
11491 of checking for a MODE_INT. Remove redundant check for whether
11492 this mode has a zero precision.
11494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11495 Alan Hayward <alan.hayward@arm.com>
11496 David Sherwood <david.sherwood@arm.com>
11498 * optabs.c (widen_leading): Change the type of the mode argument
11499 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11500 (widen_bswap): Likewise.
11501 (expand_parity): Likewise.
11502 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11503 (expand_ffs): Likewise.
11504 (epand_unop): Check for scalar integer modes before calling the
11507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11508 Alan Hayward <alan.hayward@arm.com>
11509 David Sherwood <david.sherwood@arm.com>
11511 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11513 (expand_expr_real_1): Update call accordingly.
11515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11516 Alan Hayward <alan.hayward@arm.com>
11517 David Sherwood <david.sherwood@arm.com>
11519 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11520 value_mode arguments. Use scalar_int_mode internally.
11521 (store_bit_field_1): Rename the new integer mode from imode
11522 to op0_mode and use it instead of GET_MODE (op0). Update calls
11523 to store_split_bit_field, store_bit_field_using_insv and
11524 store_fixed_bit_field.
11525 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11526 Use scalar_int_mode internally. Use a bit count rather than a mode
11527 when calculating the largest bit size for get_best_mode.
11528 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11529 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11530 Remove assertion that OP0 has a scalar integer mode.
11531 (store_split_bit_field): Add op0_mode and value_mode arguments.
11532 Update calls to extract_fixed_bit_field.
11533 (extract_bit_field_using_extv): Add an op0_mode argument.
11534 Use scalar_int_mode internally.
11535 (extract_bit_field_1): Rename the new integer mode from imode to
11536 op0_mode and use it instead of GET_MODE (op0). Update calls to
11537 extract_split_bit_field, extract_bit_field_using_extv and
11538 extract_fixed_bit_field.
11539 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11540 to extract_split_bit_field and extract_fixed_bit_field_1.
11541 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11542 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11543 on the target mode.
11544 (extract_split_bit_field): Add an op0_mode argument. Update call
11545 to extract_fixed_bit_field.
11547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11548 Alan Hayward <alan.hayward@arm.com>
11549 David Sherwood <david.sherwood@arm.com>
11551 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11552 * explow.c (hard_function_value): Likewise.
11553 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11554 convert_to_mode call outside the loop.
11555 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
11556 for the mode iterator. Require the mode specified by max_pieces
11558 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
11560 (copy_blkmode_to_reg): Likewise.
11561 (set_storage_via_setmem): Likewise.
11562 * optabs.c (prepare_cmp_insn): Likewise.
11563 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11564 * stor-layout.c (finish_bitfield_representative): Likewise.
11566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11567 Alan Hayward <alan.hayward@arm.com>
11568 David Sherwood <david.sherwood@arm.com>
11570 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
11571 * expr.c (convert_move): Use them.
11572 (convert_modes): Likewise.
11573 (store_expr_with_bounds): Likewise.
11575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11576 Alan Hayward <alan.hayward@arm.com>
11577 David Sherwood <david.sherwood@arm.com>
11579 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
11580 parameter for the mode of "x". Remove the "known_x", "known_mode"
11581 and "known_ret" arguments. Change the type of the mode argument
11582 to scalar_int_mode.
11583 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
11584 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
11585 (reg_num_sign_bit_copies_for_combine): Likewise.
11586 * rtlanal.c (nonzero_bits1): Likewise.
11587 (num_sign_bit_copies1): Likewise.
11588 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
11589 (reg_num_sign_bit_copies_general): Likewise.
11590 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
11591 (reg_nonzero_bits_general): Likewise.
11593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11594 Alan Hayward <alan.hayward@arm.com>
11595 David Sherwood <david.sherwood@arm.com>
11597 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
11598 than in subroutines. Return 1 for non-integer modes.
11599 (cached_num_sign_bit_copies): Change the type of the mode parameter
11600 to scalar_int_mode.
11601 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
11602 classes. Handle CONST_INT_P first and then check whether X also
11603 has a scalar integer mode. Check the same thing for inner registers
11604 of a SUBREG and for values that are being extended or truncated.
11606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11607 Alan Hayward <alan.hayward@arm.com>
11608 David Sherwood <david.sherwood@arm.com>
11610 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
11611 in subroutines. Return the mode mask for non-integer modes.
11612 (cached_nonzero_bits): Change the type of the mode parameter
11613 to scalar_int_mode.
11614 (nonzero_bits1): Likewise. Remove early exit for other mode
11615 classes. Handle CONST_INT_P first and then check whether X
11616 also has a scalar integer mode.
11618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11619 Alan Hayward <alan.hayward@arm.com>
11620 David Sherwood <david.sherwood@arm.com>
11622 * expr.c (widest_int_mode_for_size): Make the comment match the code.
11623 Return a scalar_int_mode and assert that the size is greater than
11625 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
11626 (op_by_pieces_d::op_by_pieces_d): Likewise.
11627 (op_by_pieces_d::run): Likewise.
11628 (can_store_by_pieces): Likewise.
11630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11631 Alan Hayward <alan.hayward@arm.com>
11632 David Sherwood <david.sherwood@arm.com>
11634 * combine.c (extract_left_shift): Add a mode argument and update
11636 (make_compound_operation_int): Change the type of the mode parameter
11637 to scalar_int_mode and update the call to extract_left_shift.
11639 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11640 Alan Hayward <alan.hayward@arm.com>
11641 David Sherwood <david.sherwood@arm.com>
11643 * combine.c (simplify_and_const_int): Change the type of the mode
11644 parameter to scalar_int_mode.
11645 (simplify_and_const_int_1): Likewise. Update recursive call.
11647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11648 Alan Hayward <alan.hayward@arm.com>
11649 David Sherwood <david.sherwood@arm.com>
11651 * combine.c (simplify_compare_const): Check that the mode is a
11652 scalar_int_mode (rather than VOIDmode) before testing its
11654 (simplify_comparison): Move COMPARISON_P handling out of the
11655 loop and restrict the latter part of the loop to scalar_int_modes.
11656 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
11657 and when considering SUBREG_REGs. Use is_int_mode instead of
11658 checking GET_MODE_CLASS against MODE_INT.
11660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11661 Alan Hayward <alan.hayward@arm.com>
11662 David Sherwood <david.sherwood@arm.com>
11664 * combine.c (try_widen_shift_mode): Move check for equal modes to...
11665 (simplify_shift_const_1): ...here. Use scalar_int_mode for
11666 shift_unit_mode and for modes involved in scalar shifts.
11668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11669 Alan Hayward <alan.hayward@arm.com>
11670 David Sherwood <david.sherwood@arm.com>
11672 * combine.c (force_int_to_mode): New function, split out from...
11673 (force_to_mode): ...here. Keep xmode up-to-date and use it
11674 instead of GET_MODE (x).
11676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11677 Alan Hayward <alan.hayward@arm.com>
11678 David Sherwood <david.sherwood@arm.com>
11680 * optabs-query.h (extraction_insn::struct_mode): Change type to
11681 opt_scalar_int_mode and update comment.
11682 (extraction_insn::field_mode): Change type to scalar_int_mode.
11683 (extraction_insn::pos_mode): Likewise.
11684 * combine.c (make_extraction): Update accordingly.
11685 * optabs-query.c (get_traditional_extraction_insn): Likewise.
11686 (get_optab_extraction_insn): Likewise.
11687 * recog.c (simplify_while_replacing): Likewise.
11688 * expmed.c (narrow_bit_field_mem): Change the type of the mode
11689 parameter to opt_scalar_int_mode.
11691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11692 Alan Hayward <alan.hayward@arm.com>
11693 David Sherwood <david.sherwood@arm.com>
11695 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
11696 to a scalar_int_mode instead of a machine_mode.
11697 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
11698 (get_best_mode): Return a boolean and use a pointer argument to store
11699 the selected mode. Replace the limit mode parameter with a bit limit.
11700 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
11701 for the values returned by bit_field_mode_iterator::next_mode.
11702 (store_bit_field): Update call to get_best_mode.
11703 (store_fixed_bit_field): Likewise.
11704 (extract_fixed_bit_field): Likewise.
11705 * expr.c (optimize_bitfield_assignment_op): Likewise.
11706 * fold-const.c (optimize_bit_field_compare): Likewise.
11707 (fold_truth_andor_1): Likewise.
11708 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
11709 Update for new type of m_mode.
11710 (get_best_mode): As above.
11712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11713 Alan Hayward <alan.hayward@arm.com>
11714 David Sherwood <david.sherwood@arm.com>
11716 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
11717 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
11718 (store_bit_field): Check is_a <scalar_int_mode> before calling
11719 strict_volatile_bitfield_p.
11720 (extract_bit_field): Likewise.
11722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11723 Alan Hayward <alan.hayward@arm.com>
11724 David Sherwood <david.sherwood@arm.com>
11726 * target.def (cstore_mode): Return a scalar_int_mode.
11727 * doc/tm.texi: Regenerate.
11728 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
11729 * targhooks.h (default_cstore_mode): Likewise.
11730 * targhooks.c (default_cstore_mode): Likewise, using a forced
11732 * expmed.c (emit_cstore): Expect the target of the cstore to be
11735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11736 Alan Hayward <alan.hayward@arm.com>
11737 David Sherwood <david.sherwood@arm.com>
11739 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11741 (niter_desc): Likewise mode.
11742 (iv_analyze): Add a mode parameter.
11744 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11745 and change its type to scalar_int_mode.
11746 * loop-iv.c: Update commentary at head of file.
11747 (iv_constant): Pass the mode paraeter before the rtx it describes
11748 and change its type to scalar_int_mode. Remove VOIDmode handling.
11749 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11750 (iv_extend): Likewise.
11751 (shorten_into_mode): Likewise.
11752 (iv_add): Use scalar_int_mode.
11753 (iv_mult): Likewise.
11754 (iv_shift): Likewise.
11755 (canonicalize_iv_subregs): Likewise.
11756 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11757 it describes and change its mode to scalar_int_mode. Also change
11758 the type of the returned inner_mode to scalar_int_mode.
11759 (get_biv_step): Likewise, turning outer_mode from a pointer
11760 into a direct parameter. Update call to get_biv_step_1.
11761 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11762 iv_constant and get_biv_step.
11763 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11764 and change its type to scalar_int_mode. Don't initialise iv->mode
11765 to VOIDmode and remove later checks for its still being VOIDmode.
11766 Update calls to iv_analyze_op and iv_analyze_expr. Check
11767 is_a <scalar_int_mode> when changing the mode under consideration.
11768 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
11769 Update call to iv_analyze_expr.
11770 (iv_analyze_op): Add a mode parameter. Reject subregs whose
11771 inner register is not also a scalar_int_mode. Update call to
11773 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
11774 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
11775 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
11776 separate mode class checks. Update calls to iv_analyze. Remove
11777 fix-up of VOIDmodes after iv_analyze_biv.
11778 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
11779 don't have a scalar_int_mode. Update call to biv_p.
11781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11782 Alan Hayward <alan.hayward@arm.com>
11783 David Sherwood <david.sherwood@arm.com>
11785 * cfgexpand.c (convert_debug_memory_address): Use
11786 as_a <scalar_int_mode>.
11787 * combine.c (expand_compound_operation): Likewise.
11788 (make_extraction): Likewise.
11789 (change_zero_ext): Likewise.
11790 (simplify_comparison): Likewise.
11791 * cse.c (cse_insn): Likewise.
11792 * dwarf2out.c (minmax_loc_descriptor): Likewise.
11793 (mem_loc_descriptor): Likewise.
11794 (loc_descriptor): Likewise.
11795 * expmed.c (init_expmed_one_mode): Likewise.
11796 (synth_mult): Likewise.
11797 (emit_store_flag_1): Likewise.
11798 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
11799 of a comparison with size.
11800 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
11801 (reduce_to_bit_field_precision): Likewise.
11802 * function.c (expand_function_end): Likewise.
11803 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
11804 * loop-doloop.c (doloop_modify): Likewise.
11805 * optabs.c (expand_binop): Likewise.
11806 (expand_unop): Likewise.
11807 (expand_copysign_absneg): Likewise.
11808 (prepare_cmp_insn): Likewise.
11809 (maybe_legitimize_operand): Likewise.
11810 * recog.c (const_scalar_int_operand): Likewise.
11811 * rtlanal.c (get_address_mode): Likewise.
11812 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11813 (simplify_cond_clz_ctz): Likewise.
11814 * tree-nested.c (get_nl_goto_field): Likewise.
11815 * tree.c (build_vector_type_for_mode): Likewise.
11816 * var-tracking.c (use_narrower_mode): Likewise.
11818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11819 Alan Hayward <alan.hayward@arm.com>
11820 David Sherwood <david.sherwood@arm.com>
11822 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11823 * builtins.c (expand_builtin_signbit): Use it.
11824 * cfgexpand.c (expand_debug_expr): Likewise.
11825 * dojump.c (do_jump): Likewise.
11826 (do_compare_and_jump): Likewise.
11827 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11828 * expmed.c (make_tree): Likewise.
11829 * expr.c (expand_expr_real_2): Likewise.
11830 (expand_expr_real_1): Likewise.
11831 (try_casesi): Likewise.
11832 * fold-const-call.c (fold_const_call_ss): Likewise.
11833 * fold-const.c (unextend): Likewise.
11834 (extract_muldiv_1): Likewise.
11835 (fold_single_bit_test): Likewise.
11836 (native_encode_int): Likewise.
11837 (native_encode_string): Likewise.
11838 (native_interpret_int): Likewise.
11839 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11840 * internal-fn.c (expand_addsub_overflow): Likewise.
11841 (expand_neg_overflow): Likewise.
11842 (expand_mul_overflow): Likewise.
11843 (expand_arith_overflow): Likewise.
11844 * match.pd: Likewise.
11845 * stor-layout.c (layout_type): Likewise.
11846 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11847 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11848 * tree-ssanames.c (get_range_info): Likewise.
11849 * tree-switch-conversion.c (array_value_type) Likewise.
11850 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11851 (vect_recog_divmod_pattern): Likewise.
11852 (vect_recog_mixed_size_cond_pattern): Likewise.
11853 * tree-vrp.c (extract_range_basic): Likewise.
11854 (simplify_float_conversion_using_ranges): Likewise.
11855 * tree.c (int_fits_type_p): Likewise.
11856 * ubsan.c (instrument_bool_enum_load): Likewise.
11857 * varasm.c (mergeable_string_section): Likewise.
11858 (narrowing_initializer_constant_valid_p): Likewise.
11859 (output_constant): Likewise.
11861 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11862 Alan Hayward <alan.hayward@arm.com>
11863 David Sherwood <david.sherwood@arm.com>
11865 * machmode.h (NARROWEST_INT_MODE): New macro.
11866 * expr.c (alignment_for_piecewise_move): Use it instead of
11867 GET_CLASS_NARROWEST_MODE (MODE_INT).
11868 (push_block): Likewise.
11869 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11871 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11873 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11874 Alan Hayward <alan.hayward@arm.com>
11875 David Sherwood <david.sherwood@arm.com>
11877 * postreload.c (move2add_valid_value_p): Change the type of the
11878 mode parameter to scalar_int_mode.
11879 (move2add_use_add2_insn): Add a mode parameter and use it instead
11881 (move2add_use_add3_insn): Likewise.
11882 (reload_cse_move2add): Update accordingly.
11884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11885 Alan Hayward <alan.hayward@arm.com>
11886 David Sherwood <david.sherwood@arm.com>
11888 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11890 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11891 * optabs.c (expand_unop): Likewise.
11893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11894 Alan Hayward <alan.hayward@arm.com>
11895 David Sherwood <david.sherwood@arm.com>
11897 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11898 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11899 (popcount_loc_descriptor): Likewise.
11900 (bswap_loc_descriptor): Likewise.
11901 (rotate_loc_descriptor): Likewise.
11902 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11903 calling the functions above.
11905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11906 Alan Hayward <alan.hayward@arm.com>
11907 David Sherwood <david.sherwood@arm.com>
11909 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11911 (try_combine): Likewise.
11912 (simplify_if_then_else): Likewise.
11913 * cse.c (cse_insn): Likewise.
11914 * dwarf2out.c (mem_loc_descriptor): Likewise.
11915 * emit-rtl.c (gen_lowpart_common): Likewise.
11916 * simplify-rtx.c (simplify_truncation): Likewise.
11917 (simplify_binary_operation_1): Likewise.
11918 (simplify_const_relational_operation): Likewise.
11919 (simplify_ternary_operation): Likewise.
11920 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11922 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11923 Alan Hayward <alan.hayward@arm.com>
11924 David Sherwood <david.sherwood@arm.com>
11926 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11927 * reload.c (push_reload): Likewise.
11928 (find_reloads): Likewise.
11930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11931 Alan Hayward <alan.hayward@arm.com>
11932 David Sherwood <david.sherwood@arm.com>
11934 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11935 (make_compound_operation_int): Likewise.
11936 (change_zero_ext): Likewise.
11937 * expr.c (convert_move): Likewise.
11938 (convert_modes): Likewise.
11939 * fwprop.c (forward_propagate_subreg): Likewise.
11940 * loop-iv.c (get_biv_step_1): Likewise.
11941 * optabs.c (widen_operand): Likewise.
11942 * postreload.c (move2add_valid_value_p): Likewise.
11943 * recog.c (simplify_while_replacing): Likewise.
11944 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11945 (simplify_binary_operation_1): Likewise. Remove redundant
11946 mode equality check.
11948 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11949 Alan Hayward <alan.hayward@arm.com>
11950 David Sherwood <david.sherwood@arm.com>
11952 * combine.c (combine_simplify_rtx): Add checks for
11953 is_a <scalar_int_mode>.
11954 (simplify_if_then_else): Likewise.
11955 (make_field_assignment): Likewise.
11956 (simplify_comparison): Likewise.
11957 * ifcvt.c (noce_try_bitop): Likewise.
11958 * loop-invariant.c (canonicalize_address_mult): Likewise.
11959 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11961 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11962 Alan Hayward <alan.hayward@arm.com>
11963 David Sherwood <david.sherwood@arm.com>
11965 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11966 is_a <scalar_int_mode> instead of != BLKmode.
11968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11969 Alan Hayward <alan.hayward@arm.com>
11970 David Sherwood <david.sherwood@arm.com>
11972 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11973 instead of != VOIDmode.
11974 * combine.c (if_then_else_cond): Likewise.
11975 (change_zero_ext): Likewise.
11976 * dwarf2out.c (mem_loc_descriptor): Likewise.
11977 (loc_descriptor): Likewise.
11978 * rtlanal.c (canonicalize_condition): Likewise.
11979 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11982 Alan Hayward <alan.hayward@arm.com>
11983 David Sherwood <david.sherwood@arm.com>
11985 * simplify-rtx.c (simplify_binary_operation_1): Use
11986 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11989 Alan Hayward <alan.hayward@arm.com>
11990 David Sherwood <david.sherwood@arm.com>
11992 * wide-int.h (int_traits<unsigned char>) New class.
11993 (int_traits<unsigned short>) Likewise.
11994 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11995 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11997 * combine.c (set_nonzero_bits_and_sign_copies): Use
11998 is_a <scalar_int_mode>.
11999 (find_split_point): Likewise.
12000 (combine_simplify_rtx): Likewise.
12001 (simplify_logical): Likewise.
12002 (expand_compound_operation): Likewise.
12003 (expand_field_assignment): Likewise.
12004 (make_compound_operation): Likewise.
12005 (extended_count): Likewise.
12006 (change_zero_ext): Likewise.
12007 (simplify_comparison): Likewise.
12008 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12009 (ucompare_loc_descriptor): Likewise.
12010 (minmax_loc_descriptor): Likewise.
12011 (mem_loc_descriptor): Likewise.
12012 (loc_descriptor): Likewise.
12013 * expmed.c (init_expmed_one_mode): Likewise.
12014 * lra-constraints.c (lra_constraint_offset): Likewise.
12015 * optabs.c (prepare_libcall_arg): Likewise.
12016 * postreload.c (move2add_note_store): Likewise.
12017 * reload.c (operands_match_p): Likewise.
12018 * rtl.h (load_extend_op): Likewise.
12019 * rtlhooks.c (gen_lowpart_general): Likewise.
12020 * simplify-rtx.c (simplify_truncation): Likewise.
12021 (simplify_unary_operation_1): Likewise.
12022 (simplify_binary_operation_1): Likewise.
12023 (simplify_const_binary_operation): Likewise.
12024 (simplify_const_relational_operation): Likewise.
12025 (simplify_subreg): Likewise.
12026 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12027 * var-tracking.c (adjust_mems): Likewise.
12028 (prepare_call_arguments): Likewise.
12030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12031 Alan Hayward <alan.hayward@arm.com>
12032 David Sherwood <david.sherwood@arm.com>
12034 * machmode.h (is_int_mode): New fuction.
12035 * combine.c (find_split_point): Use it.
12036 (combine_simplify_rtx): Likewise.
12037 (simplify_if_then_else): Likewise.
12038 (simplify_set): Likewise.
12039 (simplify_shift_const_1): Likewise.
12040 (simplify_comparison): Likewise.
12041 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12042 * cse.c (notreg_cost): Likewise.
12043 (cse_insn): Likewise.
12044 * cselib.c (cselib_lookup_1): Likewise.
12045 * dojump.c (do_jump_1): Likewise.
12046 (do_compare_rtx_and_jump): Likewise.
12047 * dse.c (get_call_args): Likewise.
12048 * dwarf2out.c (rtl_for_decl_init): Likewise.
12049 (native_encode_initializer): Likewise.
12050 * expmed.c (emit_store_flag_1): Likewise.
12051 (emit_store_flag): Likewise.
12052 * expr.c (convert_modes): Likewise.
12053 (store_field): Likewise.
12054 (expand_expr_real_1): Likewise.
12055 * fold-const.c (fold_read_from_constant_string): Likewise.
12056 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12057 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12058 * optabs.c (expand_binop): Likewise.
12059 (expand_unop): Likewise.
12060 (expand_abs_nojump): Likewise.
12061 (expand_one_cmpl_abs_nojump): Likewise.
12062 * simplify-rtx.c (mode_signbit_p): Likewise.
12063 (val_signbit_p): Likewise.
12064 (val_signbit_known_set_p): Likewise.
12065 (val_signbit_known_clear_p): Likewise.
12066 (simplify_relational_operation_1): Likewise.
12067 * tree.c (vector_type_mode): Likewise.
12069 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12070 Alan Hayward <alan.hayward@arm.com>
12071 David Sherwood <david.sherwood@arm.com>
12073 * machmode.h (smallest_mode_for_size): Fix formatting.
12074 (smallest_int_mode_for_size): New function.
12075 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12076 instead of smallest_mode_for_size.
12077 * combine.c (make_extraction): Likewise.
12078 * config/arc/arc.c (arc_expand_movmem): Likewise.
12079 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12080 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12081 * config/s390/s390.c (s390_expand_insv): Likewise.
12082 * config/sparc/sparc.c (assign_int_registers): Likewise.
12083 * config/spu/spu.c (spu_function_value): Likewise.
12084 (spu_function_arg): Likewise.
12085 * coverage.c (get_gcov_type): Likewise.
12086 (get_gcov_unsigned_t): Likewise.
12087 * dse.c (find_shift_sequence): Likewise.
12088 * expmed.c (store_bit_field_1): Likewise.
12089 * expr.c (convert_move): Likewise.
12090 (store_field): Likewise.
12091 * internal-fn.c (expand_arith_overflow): Likewise.
12092 * optabs-query.c (get_best_extraction_insn): Likewise.
12093 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12094 * stor-layout.c (layout_type): Likewise.
12095 (initialize_sizetypes): Likewise.
12096 * targhooks.c (default_get_mask_mode): Likewise.
12097 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12099 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12100 Alan Hayward <alan.hayward@arm.com>
12101 David Sherwood <david.sherwood@arm.com>
12103 * machmode.h (opt_mode::else_blk): New function.
12104 (int_mode_for_mode): Declare.
12105 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12106 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12108 * cfgexpand.c (expand_debug_expr): Likewise.
12109 * combine.c (gen_lowpart_or_truncate): Likewise.
12110 (gen_lowpart_for_combine): Likewise.
12111 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12112 * config/avr/avr.c (avr_to_int_mode): Likewise.
12113 (avr_out_plus_1): Likewise.
12114 (avr_out_plus): Likewise.
12115 (avr_out_round): Likewise.
12116 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12117 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12118 (s390_expand_vcond): Likewise.
12119 * config/spu/spu.c (spu_split_immediate): Likewise.
12120 (spu_expand_mov): Likewise.
12121 * dse.c (get_stored_val): Likewise.
12122 * expmed.c (store_bit_field_1): Likewise.
12123 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12125 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12126 (extract_low_bits): Likewise.
12127 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12128 handling rather than repeating the check.
12129 (emit_group_store): Likewise.
12130 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12131 * optabs.c (expand_absneg_bit): Likewise.
12132 (expand_copysign_absneg): Likewise.
12133 (expand_copysign_bit): Likewise.
12134 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12135 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12136 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12137 * var-tracking.c (prepare_call_arguments): Likewise.
12138 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12139 int_mode_for_mode instead of mode_for_size.
12140 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12143 Alan Hayward <alan.hayward@arm.com>
12144 David Sherwood <david.sherwood@arm.com>
12146 * machmode.h (int_mode_for_size): New function.
12147 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12148 instead of mode_for_size.
12149 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12151 * combine.c (expand_field_assignment): Use int_mode_for_size
12152 instead of mode_for_size.
12153 (make_extraction): Likewise.
12154 (simplify_shift_const_1): Likewise.
12155 (simplify_comparison): Likewise.
12156 * dojump.c (do_jump): Likewise.
12157 * dwarf2out.c (mem_loc_descriptor): Likewise.
12158 * emit-rtl.c (init_derived_machine_modes): Likewise.
12159 * expmed.c (flip_storage_order): Likewise.
12160 (convert_extracted_bit_field): Likewise.
12161 * expr.c (copy_blkmode_from_reg): Likewise.
12162 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12163 * internal-fn.c (expand_mul_overflow): Likewise.
12164 * lower-subreg.c (simple_move): Likewise.
12165 * optabs-libfuncs.c (init_optabs): Likewise.
12166 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12167 * tree.c (vector_type_mode): Likewise.
12168 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12169 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12170 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12171 * tree-vect-stmts.c (vectorizable_load): Likewise.
12172 (vectorizable_store): Likewise.
12174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12175 Alan Hayward <alan.hayward@arm.com>
12176 David Sherwood <david.sherwood@arm.com>
12178 * coretypes.h (pod_mode): New type.
12179 (scalar_int_mode_pod): New typedef.
12180 * machmode.h (pod_mode): New class.
12181 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12182 * genmodes.c (emit_mode_int_n): Update accordingly.
12183 * lower-subreg.h (target_lower_subreg): Change type to
12184 scalar_int_mode_pod.
12185 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12186 scalar_int_mode_pod.
12188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12189 Alan Hayward <alan.hayward@arm.com>
12190 David Sherwood <david.sherwood@arm.com>
12192 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12193 machine_mode to scalar_int_mode.
12194 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12195 (rs6000_option_override_internal): Remove cast to int.
12196 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12197 machine_mode to scalar_int_mode.
12198 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12199 (rs6000_option_override_internal): Remove cast to int.
12200 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12201 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12203 * config/s390/s390.c (s390_expand_builtin): Likewise.
12204 * coretypes.h (scalar_int_mode): New type.
12205 (opt_scalar_int_mode): New typedef.
12206 * machmode.h (scalar_int_mode): New class.
12207 (scalar_int_mode::includes_p): New function.
12208 (byte_mode): Change type to scalar_int_mode.
12209 (word_mode): Likewise.
12210 (ptr_mode): Likewise.
12211 * emit-rtl.c (byte_mode): Likewise.
12212 (word_mode): Likewise.
12213 (ptr_mode): Likewise.
12214 (init_derived_machine_modes): Update accordingly.
12215 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12216 and MODE_PARTIAL_INT.
12217 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12218 opt_scalar_int_mode.
12220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12221 Alan Hayward <alan.hayward@arm.com>
12222 David Sherwood <david.sherwood@arm.com>
12224 * target.def (libgcc_floating_mode_supported_p): Take a
12226 * doc/tm.texi: Regenerate.
12227 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12229 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12230 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12233 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12234 Alan Hayward <alan.hayward@arm.com>
12235 David Sherwood <david.sherwood@arm.com>
12237 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12238 * doc/tm.texi: Regenerate.
12239 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12240 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12241 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12242 * targhooks.h (default_floatn_mode): Likewise.
12243 * targhooks.c (default_floatn_mode): Likewise.
12244 * tree.c (build_common_tree_nodes): Update accordingly.
12246 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12247 Alan Hayward <alan.hayward@arm.com>
12248 David Sherwood <david.sherwood@arm.com>
12250 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12251 (mode_iterator::iterate_p): Likewise.
12252 (mode_iterator::get_wider): Likewise.
12253 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12256 Alan Hayward <alan.hayward@arm.com>
12257 David Sherwood <david.sherwood@arm.com>
12259 * coretypes.h (opt_scalar_float_mode): New typedef.
12260 * machmode.h (float_mode_for_size): New function.
12261 * emit-rtl.c (double_mode): Delete.
12262 (init_emit_once): Use float_mode_for_size.
12263 * stor-layout.c (layout_type): Likewise.
12264 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12267 Alan Hayward <alan.hayward@arm.com>
12268 David Sherwood <david.sherwood@arm.com>
12270 * output.h (assemble_real): Take a scalar_float_mode.
12271 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12272 * config/arm/arm.md (consttable_4): Likewise.
12273 (consttable_8): Likewise.
12274 (consttable_16): Likewise.
12275 * config/mips/mips.md (consttable_float): Likewise.
12276 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12277 * varasm.c (assemble_real): Take a scalar_float_mode.
12278 (output_constant_pool_2): Update accordingly.
12279 (output_constant): Likewise.
12281 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12282 Alan Hayward <alan.hayward@arm.com>
12283 David Sherwood <david.sherwood@arm.com>
12285 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12286 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12287 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12288 (native_encode_real): Likewise.
12289 (native_interpret_real): Likewise.
12290 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12291 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12293 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12294 Alan Hayward <alan.hayward@arm.com>
12295 David Sherwood <david.sherwood@arm.com>
12297 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12298 <scalar_float_mode>. Simplify.
12299 (gen_extend_conv_libfunc): Likewise.
12301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12302 Alan Hayward <alan.hayward@arm.com>
12303 David Sherwood <david.sherwood@arm.com>
12305 * coretypes.h (scalar_float_mode): New type.
12306 * machmode.h (mode_traits::from_int): Use machine_mode if
12307 USE_ENUM_MODES is defined.
12308 (is_a): New function.
12310 (dyn_cast): Likewise.
12311 (scalar_float_mode): New class.
12312 (scalar_float_mode::includes_p): New function.
12313 (is_float_mode): Likewise.
12314 * gdbhooks.py (MachineModePrinter): New class.
12315 (build_pretty_printer): Use it for scalar_float_mode.
12316 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12317 (format_helper::format_helper): Turn into a template.
12318 * genmodes.c (get_mode_class): New function.
12319 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12320 or machine_mode if none.
12321 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12322 as_a <scalar_float_mode>.
12323 * dwarf2out.c (mem_loc_descriptor): Likewise.
12324 (insert_float): Likewise.
12325 (add_const_value_attribute): Likewise.
12326 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12327 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12328 (expand_unop): Update accordingly.
12329 (expand_abs_nojump): Likewise.
12330 (expand_copysign_absneg): Take a scalar_float_mode.
12331 (expand_copysign_bit): Likewise.
12332 (expand_copysign): Update accordingly.
12334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12335 Alan Hayward <alan.hayward@arm.com>
12336 David Sherwood <david.sherwood@arm.com>
12338 * coretypes.h (opt_mode): New class.
12339 * machmode.h (opt_mode): Likewise.
12340 (opt_mode::else_void): New function.
12341 (opt_mode::require): Likewise.
12342 (opt_mode::exists): Likewise.
12343 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12344 (GET_MODE_2XWIDER_MODE): Likewise.
12345 (mode_iterator::get_wider): Update accordingly.
12346 (mode_iterator::get_2xwider): Likewise.
12347 (mode_iterator::get_known_wider): Likewise, turning into a template.
12348 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12349 forcing a wider mode to exist.
12350 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12351 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12352 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12353 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12354 * lower-subreg.c (init_lower_subreg): Likewise.
12355 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12356 on the final iteration.
12357 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12358 a wider mode exists before asking for a move pattern.
12359 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12360 forcing a wider mode to exist.
12361 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12362 returning false if no such mode exists.
12363 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12364 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12365 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12366 Avoid checking for a MODE_INT if we already know the mode is not a
12368 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12369 forcing a wider mode to exist.
12370 (expmed_mult_highpart_optab): Likewise.
12371 (expmed_mult_highpart): Likewise.
12372 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12374 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12375 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12376 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12377 * internal-fn.c (expand_mul_overflow): Update use of
12378 GET_MODE_2XWIDER_MODE.
12379 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12380 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12381 GET_MODE_WIDER_MODE.
12382 (convert_plusminus_to_widen): Likewise.
12383 * tree-switch-conversion.c (array_value_type): Likewise.
12384 * var-tracking.c (emit_note_insn_var_location): Likewise.
12385 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12386 Return false inside rather than outside the loop if no wider mode
12388 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12389 and GET_MODE_2XWIDER_MODE
12390 (can_compare_p): Use else_void.
12391 * gdbhooks.py (OptMachineModePrinter): New class.
12392 (build_pretty_printer): Use it for opt_mode.
12394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12395 Alan Hayward <alan.hayward@arm.com>
12396 David Sherwood <david.sherwood@arm.com>
12398 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12399 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12402 Alan Hayward <alan.hayward@arm.com>
12403 David Sherwood <david.sherwood@arm.com>
12405 * machmode.h (mode_traits): New structure.
12406 (get_narrowest_mode): New function.
12407 (mode_iterator::start): Likewise.
12408 (mode_iterator::iterate_p): Likewise.
12409 (mode_iterator::get_wider): Likewise.
12410 (mode_iterator::get_known_wider): Likewise.
12411 (mode_iterator::get_2xwider): Likewise.
12412 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12413 (FOR_EACH_MODE): Likewise.
12414 (FOR_EACH_MODE_FROM): Likewise.
12415 (FOR_EACH_MODE_UNTIL): Likewise.
12416 (FOR_EACH_WIDER_MODE): Likewise.
12417 (FOR_EACH_2XWIDER_MODE): Likewise.
12418 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12419 * combine.c (simplify_comparison): Likewise
12420 * config/i386/i386.c (type_natural_mode): Likewise.
12421 * cse.c (cse_insn): Likewise.
12422 * dse.c (find_shift_sequence): Likewise.
12423 * emit-rtl.c (init_derived_machine_modes): Likewise.
12424 (init_emit_once): Likewise.
12425 * explow.c (hard_function_value): Likewise.
12426 * expmed.c (extract_fixed_bit_field_1): Likewise.
12427 (extract_bit_field_1): Likewise.
12428 (expand_divmod): Likewise.
12429 (emit_store_flag_1): Likewise.
12430 * expr.c (init_expr_target): Likewise.
12431 (convert_move): Likewise.
12432 (alignment_for_piecewise_move): Likewise.
12433 (widest_int_mode_for_size): Likewise.
12434 (emit_block_move_via_movmem): Likewise.
12435 (copy_blkmode_to_reg): Likewise.
12436 (set_storage_via_setmem): Likewise.
12437 (compress_float_constant): Likewise.
12438 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12439 * optabs-query.c (get_best_extraction_insn): Likewise.
12440 * optabs.c (expand_binop): Likewise.
12441 (expand_twoval_unop): Likewise.
12442 (expand_twoval_binop): Likewise.
12443 (widen_leading): Likewise.
12444 (widen_bswap): Likewise.
12445 (expand_parity): Likewise.
12446 (expand_unop): Likewise.
12447 (prepare_cmp_insn): Likewise.
12448 (prepare_float_lib_cmp): Likewise.
12449 (expand_float): Likewise.
12450 (expand_fix): Likewise.
12451 (expand_sfix_optab): Likewise.
12452 * postreload.c (move2add_use_add2_insn): Likewise.
12453 * reg-stack.c (reg_to_stack): Likewise.
12454 * reginfo.c (choose_hard_reg_mode): Likewise.
12455 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12456 * stor-layout.c (mode_for_size): Likewise.
12457 (smallest_mode_for_size): Likewise.
12458 (mode_for_vector): Likewise.
12459 (finish_bitfield_representative): Likewise.
12460 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12461 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12462 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12463 * var-tracking.c (prepare_call_arguments): Likewise.
12465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12466 Alan Hayward <alan.hayward@arm.com>
12467 David Sherwood <david.sherwood@arm.com>
12469 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12470 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12471 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12472 * machmode.h (mode_size): Move earlier in file.
12473 (mode_precision): Likewise.
12474 (mode_inner): Likewise.
12475 (mode_nunits): Likewise.
12476 (mode_unit_size): Likewise.
12477 (unit_unit_precision): Likewise.
12478 (mode_wider): Likewise.
12479 (mode_2xwider): Likewise.
12480 (machine_mode): New class.
12481 (mode_to_bytes): New function.
12482 (mode_to_bits): Likewise.
12483 (mode_to_precision): Likewise.
12484 (mode_to_inner): Likewise.
12485 (mode_to_unit_size): Likewise.
12486 (mode_to_unit_precision): Likewise.
12487 (mode_to_nunits): Likewise.
12488 (GET_MODE_SIZE): Use mode_to_bytes.
12489 (GET_MODE_BITSIZE): Use mode_to_bits.
12490 (GET_MODE_PRECISION): Use mode_to_precision.
12491 (GET_MODE_INNER): Use mode_to_inner.
12492 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12493 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12494 (GET_MODE_NUNITS): Use mode_to_nunits.
12495 * system.h (ALWAYS_INLINE): New macro.
12496 * config/powerpcspe/powerpcspe-c.c
12497 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12498 int for arg1_mode and arg2_mode.
12500 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12501 Alan Hayward <alan.hayward@arm.com>
12502 David Sherwood <david.sherwood@arm.com>
12504 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12505 Prefix mode names with E_ in case statements.
12506 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12507 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12508 (aarch64_split_simd_move): Likewise.
12509 (aarch64_gen_storewb_pair): Likewise.
12510 (aarch64_gen_loadwb_pair): Likewise.
12511 (aarch64_gen_store_pair): Likewise.
12512 (aarch64_gen_load_pair): Likewise.
12513 (aarch64_get_condition_code_1): Likewise.
12514 (aarch64_constant_pool_reload_icode): Likewise.
12515 (get_rsqrte_type): Likewise.
12516 (get_rsqrts_type): Likewise.
12517 (get_recpe_type): Likewise.
12518 (get_recps_type): Likewise.
12519 (aarch64_gimplify_va_arg_expr): Likewise.
12520 (aarch64_simd_container_mode): Likewise.
12521 (aarch64_emit_load_exclusive): Likewise.
12522 (aarch64_emit_store_exclusive): Likewise.
12523 (aarch64_expand_compare_and_swap): Likewise.
12524 (aarch64_gen_atomic_cas): Likewise.
12525 (aarch64_emit_bic): Likewise.
12526 (aarch64_emit_atomic_swap): Likewise.
12527 (aarch64_emit_atomic_load_op): Likewise.
12528 (aarch64_evpc_trn): Likewise.
12529 (aarch64_evpc_uzp): Likewise.
12530 (aarch64_evpc_zip): Likewise.
12531 (aarch64_evpc_ext): Likewise.
12532 (aarch64_evpc_rev): Likewise.
12533 (aarch64_evpc_dup): Likewise.
12534 (aarch64_gen_ccmp_first): Likewise.
12535 (aarch64_gen_ccmp_next): Likewise.
12536 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12537 (alpha_emit_xfloating_libcall): Likewise.
12538 (emit_insxl): Likewise.
12539 (alpha_arg_type): Likewise.
12540 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12541 (arc_preferred_simd_mode): Likewise.
12542 (arc_secondary_reload): Likewise.
12543 (get_arc_condition_code): Likewise.
12544 (arc_print_operand): Likewise.
12545 (arc_legitimate_constant_p): Likewise.
12546 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12547 * config/arc/arc.md (casesi_load): Likewise.
12548 (casesi_compact_jump): Likewise.
12549 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12550 (cc_use_register): Likewise.
12551 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12552 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12553 (arm_init_iwmmxt_builtins): Likewise.
12554 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12555 (neon_expand_vector_init): Likewise.
12556 (arm_attr_length_move_neon): Likewise.
12557 (maybe_get_arm_condition_code): Likewise.
12558 (arm_emit_vector_const): Likewise.
12559 (arm_preferred_simd_mode): Likewise.
12560 (arm_output_iwmmxt_tinsr): Likewise.
12561 (thumb1_output_casesi): Likewise.
12562 (thumb2_output_casesi): Likewise.
12563 (arm_emit_load_exclusive): Likewise.
12564 (arm_emit_store_exclusive): Likewise.
12565 (arm_expand_compare_and_swap): Likewise.
12566 (arm_evpc_neon_vuzp): Likewise.
12567 (arm_evpc_neon_vzip): Likewise.
12568 (arm_evpc_neon_vrev): Likewise.
12569 (arm_evpc_neon_vtrn): Likewise.
12570 (arm_evpc_neon_vext): Likewise.
12571 (arm_validize_comparison): Likewise.
12572 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
12573 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
12574 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
12575 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
12576 (c6x_preferred_simd_mode): Likewise.
12577 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
12578 (epiphany_rtx_costs): Likewise.
12579 * config/epiphany/predicates.md (proper_comparison_operator):
12581 * config/frv/frv.c (condexec_memory_operand): Likewise.
12582 (frv_emit_move): Likewise.
12583 (output_move_single): Likewise.
12584 (output_condmove_single): Likewise.
12585 (frv_hard_regno_mode_ok): Likewise.
12586 (frv_matching_accg_mode): Likewise.
12587 * config/h8300/h8300.c (split_adds_subs): Likewise.
12588 (h8300_rtx_costs): Likewise.
12589 (h8300_print_operand): Likewise.
12590 (compute_mov_length): Likewise.
12591 (output_logical_op): Likewise.
12592 (compute_logical_op_length): Likewise.
12593 (compute_logical_op_cc): Likewise.
12594 (h8300_shift_needs_scratch_p): Likewise.
12595 (output_a_shift): Likewise.
12596 (compute_a_shift_length): Likewise.
12597 (compute_a_shift_cc): Likewise.
12598 (expand_a_rotate): Likewise.
12599 (output_a_rotate): Likewise.
12600 * config/i386/i386.c (classify_argument): Likewise.
12601 (function_arg_advance_32): Likewise.
12602 (function_arg_32): Likewise.
12603 (function_arg_64): Likewise.
12604 (function_value_64): Likewise.
12605 (ix86_gimplify_va_arg): Likewise.
12606 (ix86_legitimate_constant_p): Likewise.
12607 (put_condition_code): Likewise.
12608 (split_double_mode): Likewise.
12609 (ix86_avx256_split_vector_move_misalign): Likewise.
12610 (ix86_expand_vector_logical_operator): Likewise.
12611 (ix86_split_idivmod): Likewise.
12612 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
12613 (ix86_build_const_vector): Likewise.
12614 (ix86_build_signbit_mask): Likewise.
12615 (ix86_match_ccmode): Likewise.
12616 (ix86_cc_modes_compatible): Likewise.
12617 (ix86_expand_branch): Likewise.
12618 (ix86_expand_sse_cmp): Likewise.
12619 (ix86_expand_sse_movcc): Likewise.
12620 (ix86_expand_int_sse_cmp): Likewise.
12621 (ix86_expand_vec_perm_vpermi2): Likewise.
12622 (ix86_expand_vec_perm): Likewise.
12623 (ix86_expand_sse_unpack): Likewise.
12624 (ix86_expand_int_addcc): Likewise.
12625 (ix86_split_to_parts): Likewise.
12626 (ix86_vectorize_builtin_gather): Likewise.
12627 (ix86_vectorize_builtin_scatter): Likewise.
12628 (avx_vpermilp_parallel): Likewise.
12629 (inline_memory_move_cost): Likewise.
12630 (ix86_tieable_integer_mode_p): Likewise.
12631 (x86_maybe_negate_const_int): Likewise.
12632 (ix86_expand_vector_init_duplicate): Likewise.
12633 (ix86_expand_vector_init_one_nonzero): Likewise.
12634 (ix86_expand_vector_init_one_var): Likewise.
12635 (ix86_expand_vector_init_concat): Likewise.
12636 (ix86_expand_vector_init_interleave): Likewise.
12637 (ix86_expand_vector_init_general): Likewise.
12638 (ix86_expand_vector_set): Likewise.
12639 (ix86_expand_vector_extract): Likewise.
12640 (emit_reduc_half): Likewise.
12641 (ix86_emit_i387_round): Likewise.
12642 (ix86_mangle_type): Likewise.
12643 (ix86_expand_round_sse4): Likewise.
12644 (expand_vec_perm_blend): Likewise.
12645 (canonicalize_vector_int_perm): Likewise.
12646 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
12647 (expand_vec_perm_1): Likewise.
12648 (expand_vec_perm_interleave3): Likewise.
12649 (expand_vec_perm_even_odd_pack): Likewise.
12650 (expand_vec_perm_even_odd_1): Likewise.
12651 (expand_vec_perm_broadcast_1): Likewise.
12652 (ix86_vectorize_vec_perm_const_ok): Likewise.
12653 (ix86_expand_vecop_qihi): Likewise.
12654 (ix86_expand_mul_widen_hilo): Likewise.
12655 (ix86_expand_sse2_abs): Likewise.
12656 (ix86_expand_pextr): Likewise.
12657 (ix86_expand_pinsr): Likewise.
12658 (ix86_preferred_simd_mode): Likewise.
12659 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
12660 * config/i386/sse.md (*andnot<mode>3): Likewise.
12661 (<mask_codefor><code><mode>3<mask_name>): Likewise.
12662 (*<code><mode>3): Likewise.
12663 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
12664 (ia64_expand_atomic_op): Likewise.
12665 (ia64_arg_type): Likewise.
12666 (ia64_mode_to_int): Likewise.
12667 (ia64_scalar_mode_supported_p): Likewise.
12668 (ia64_vector_mode_supported_p): Likewise.
12669 (expand_vec_perm_broadcast): Likewise.
12670 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
12671 (iq2000_function_arg_advance): Likewise.
12672 (iq2000_function_arg): Likewise.
12673 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
12674 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
12675 (m68k_libcall_value): Likewise.
12676 (m68k_function_value): Likewise.
12677 (sched_attr_op_type): Likewise.
12678 * config/mcore/mcore.c (mcore_output_move): Likewise.
12679 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
12681 (microblaze_function_arg): Likewise.
12682 * config/mips/mips.c (mips16_build_call_stub): Likewise.
12683 (mips_print_operand): Likewise.
12684 (mips_mode_ok_for_mov_fmt_p): Likewise.
12685 (mips_vector_mode_supported_p): Likewise.
12686 (mips_preferred_simd_mode): Likewise.
12687 (mips_expand_vpc_loongson_even_odd): Likewise.
12688 (mips_expand_vec_unpack): Likewise.
12689 (mips_expand_vi_broadcast): Likewise.
12690 (mips_expand_vector_init): Likewise.
12691 (mips_expand_vec_reduc): Likewise.
12692 (mips_expand_msa_cmp): Likewise.
12693 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
12694 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
12695 (cc_flags_for_mode): Likewise.
12696 * config/msp430/msp430.c (msp430_print_operand): Likewise.
12697 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
12698 (nds32_output_casesi_pc_relative): Likewise.
12699 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12700 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
12701 (nvptx_gen_unpack): Likewise.
12702 (nvptx_gen_pack): Likewise.
12703 (nvptx_gen_shuffle): Likewise.
12704 (nvptx_gen_wcast): Likewise.
12705 (nvptx_preferred_simd_mode): Likewise.
12706 * config/pa/pa.c (pa_secondary_reload): Likewise.
12707 * config/pa/predicates.md (base14_operand): Likewise.
12708 * config/powerpcspe/powerpcspe-c.c
12709 (altivec_resolve_overloaded_builtin): Likewise.
12710 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
12712 (rs6000_preferred_simd_mode): Likewise.
12713 (output_vec_const_move): Likewise.
12714 (rs6000_expand_vector_extract): Likewise.
12715 (rs6000_split_vec_extract_var): Likewise.
12716 (reg_offset_addressing_ok_p): Likewise.
12717 (rs6000_legitimate_offset_address_p): Likewise.
12718 (rs6000_legitimize_address): Likewise.
12719 (rs6000_emit_set_const): Likewise.
12720 (rs6000_const_vec): Likewise.
12721 (rs6000_emit_move): Likewise.
12722 (spe_build_register_parallel): Likewise.
12723 (rs6000_darwin64_record_arg_recurse): Likewise.
12724 (swap_selector_for_mode): Likewise.
12725 (spe_init_builtins): Likewise.
12726 (paired_init_builtins): Likewise.
12727 (altivec_init_builtins): Likewise.
12728 (do_load_for_compare): Likewise.
12729 (rs6000_generate_compare): Likewise.
12730 (rs6000_expand_float128_convert): Likewise.
12731 (emit_load_locked): Likewise.
12732 (emit_store_conditional): Likewise.
12733 (rs6000_output_function_epilogue): Likewise.
12734 (rs6000_handle_altivec_attribute): Likewise.
12735 (rs6000_function_value): Likewise.
12736 (emit_fusion_gpr_load): Likewise.
12737 (emit_fusion_p9_load): Likewise.
12738 (emit_fusion_p9_store): Likewise.
12739 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12740 (fusion_gpr_mem_load): Likewise.
12741 (fusion_addis_mem_combo_load): Likewise.
12742 (fusion_addis_mem_combo_store): Likewise.
12743 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
12744 (fusion_gpr_mem_load): Likewise.
12745 (fusion_addis_mem_combo_load): Likewise.
12746 (fusion_addis_mem_combo_store): Likewise.
12747 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12749 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12750 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12751 (rs6000_preferred_simd_mode): Likewise.
12752 (output_vec_const_move): Likewise.
12753 (rs6000_expand_vector_extract): Likewise.
12754 (rs6000_split_vec_extract_var): Likewise.
12755 (reg_offset_addressing_ok_p): Likewise.
12756 (rs6000_legitimate_offset_address_p): Likewise.
12757 (rs6000_legitimize_address): Likewise.
12758 (rs6000_emit_set_const): Likewise.
12759 (rs6000_const_vec): Likewise.
12760 (rs6000_emit_move): Likewise.
12761 (rs6000_darwin64_record_arg_recurse): Likewise.
12762 (swap_selector_for_mode): Likewise.
12763 (paired_init_builtins): Likewise.
12764 (altivec_init_builtins): Likewise.
12765 (rs6000_expand_float128_convert): Likewise.
12766 (emit_load_locked): Likewise.
12767 (emit_store_conditional): Likewise.
12768 (rs6000_output_function_epilogue): Likewise.
12769 (rs6000_handle_altivec_attribute): Likewise.
12770 (rs6000_function_value): Likewise.
12771 (emit_fusion_gpr_load): Likewise.
12772 (emit_fusion_p9_load): Likewise.
12773 (emit_fusion_p9_store): Likewise.
12774 * config/rx/rx.c (rx_gen_move_template): Likewise.
12775 (flags_from_mode): Likewise.
12776 * config/s390/predicates.md (s390_alc_comparison): Likewise.
12777 (s390_slb_comparison): Likewise.
12778 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
12779 (s390_vector_mode_supported_p): Likewise.
12780 (s390_cc_modes_compatible): Likewise.
12781 (s390_match_ccmode_set): Likewise.
12782 (s390_canonicalize_comparison): Likewise.
12783 (s390_emit_compare_and_swap): Likewise.
12784 (s390_branch_condition_mask): Likewise.
12785 (s390_rtx_costs): Likewise.
12786 (s390_secondary_reload): Likewise.
12787 (__SECONDARY_RELOAD_CASE): Likewise.
12788 (s390_expand_cs): Likewise.
12789 (s390_preferred_simd_mode): Likewise.
12790 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
12791 * config/sh/sh.c (sh_print_operand): Likewise.
12792 (dump_table): Likewise.
12793 (sh_secondary_reload): Likewise.
12794 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12795 * config/sh/sh.md (casesi_worker_1): Likewise.
12796 (casesi_worker_2): Likewise.
12797 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
12798 (fcc_comparison_operator): Likewise.
12799 * config/sparc/sparc.c (sparc_expand_move): Likewise.
12800 (emit_soft_tfmode_cvt): Likewise.
12801 (sparc_preferred_simd_mode): Likewise.
12802 (output_cbranch): Likewise.
12803 (sparc_print_operand): Likewise.
12804 (sparc_expand_vec_perm_bmask): Likewise.
12805 (vector_init_bshuffle): Likewise.
12806 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12807 (spu_vector_mode_supported_p): Likewise.
12808 (spu_expand_insv): Likewise.
12809 (spu_emit_branch_or_set): Likewise.
12810 (spu_handle_vector_attribute): Likewise.
12811 (spu_builtin_splats): Likewise.
12812 (spu_builtin_extract): Likewise.
12813 (spu_builtin_promote): Likewise.
12814 (spu_expand_sign_extend): Likewise.
12815 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12816 (tilegx_simd_int): Likewise.
12817 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
12818 (tilepro_simd_int): Likewise.
12819 * config/v850/v850.c (const_double_split): Likewise.
12820 (v850_print_operand): Likewise.
12821 (ep_memory_offset): Likewise.
12822 * config/vax/vax.c (vax_rtx_costs): Likewise.
12823 (vax_output_int_move): Likewise.
12824 (vax_output_int_add): Likewise.
12825 (vax_output_int_subtract): Likewise.
12826 * config/visium/predicates.md (visium_branch_operator): Likewise.
12827 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12828 (visium_print_operand_address): Likewise.
12829 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12830 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12831 (xtensa_expand_conditional_branch): Likewise.
12832 (xtensa_copy_incoming_a7): Likewise.
12833 (xtensa_output_literal): Likewise.
12834 * dfp.c (decimal_real_maxval): Likewise.
12835 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12838 Alan Hayward <alan.hayward@arm.com>
12839 David Sherwood <david.sherwood@arm.com>
12841 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12842 (mode_nunits_inline): Likewise.
12843 (mode_inner_inline): Likewise.
12844 (mode_unit_size_inline): Likewise.
12845 (mode_unit_precision_inline): Likewise.
12846 (emit_insn_modes_h): Likewise. Also emit a #define of the
12848 (emit_mode_wider): Add an E_ prefix to mode names.
12849 (emit_mode_complex): Likewise.
12850 (emit_mode_inner): Likewise.
12851 (emit_mode_adjustments): Likewise.
12852 (emit_mode_int_n): Likewise.
12853 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12854 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12855 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12856 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12857 (CRC32_BUILTIN, ENTRY): Likewise.
12858 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12859 (aarch64_pop_regs): Likewise.
12860 (aarch64_process_components): Likewise.
12861 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12862 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12863 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12864 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12865 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12866 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12868 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12869 (pcmpistr): Likewise.
12870 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12871 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12872 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12874 * config/rl78/rl78.c (mduc_regs): Likewise.
12875 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12876 (htm_expand_builtin): Likewise.
12877 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12878 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12879 * config/xtensa/xtensa.c (print_operand): Likewise.
12880 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12881 (NUM_MODE_VECTOR_INT): Likewise.
12882 * genoutput.c (null_operand): Likewise.
12883 (output_operand_data): Likewise.
12884 * genrecog.c (print_parameter_value): Likewise.
12885 * lra.c (debug_operand_data): Likewise.
12887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12889 * dwarf2out.c (scompare_loc_descriptor_wide)
12890 (scompare_loc_descriptor_narrow): New functions, split out from...
12891 (scompare_loc_descriptor): ...here.
12892 * expmed.c (emit_store_flag_int): New function, split out from...
12893 (emit_store_flag): ...here.
12895 2017-08-30 Richard Biener <rguenther@suse.de>
12897 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12898 (dwarf2out_early_finish): Move setting of AT_pubnames from
12899 early debug output to early finish.
12901 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12903 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12904 and -mdata-region to the assembler.
12906 2017-08-30 Richard Biener <rguenther@suse.de>
12908 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12910 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12911 (dwarf2out_early_global_decl): Only generate a DIE for the
12912 abstract origin if it doesn't already exist or is a declaration DIE.
12913 (resolve_addr): Do not add the linkage name twice when
12914 generating a stub DIE for the DW_TAG_GNU_call_site target.
12916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12918 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12919 Use machine_mode rather than int for arg1_mode.
12921 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12924 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12925 that the second argument of the built-in functions to unpack
12926 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12927 switch statement instead a lot of if statements.
12928 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12929 Allow 64-bit values to be in Altivec registers as well as
12930 traditional floating point registers.
12931 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12933 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12935 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12936 MAX_REGS_PER_ADDRESS == 1.
12938 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12940 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12941 * config/i386/i386.c (ix86_option_override_internal): Simplify
12942 setting of opts->x_flag_entry.
12944 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12945 Jakub Jelinek <jakub@redhat.com>
12946 Richard Biener <rguenther@suse.de>
12948 PR tree-optimization/81503
12949 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12950 folded constant fits in the target type; reorder tests for clarity.
12952 2017-08-29 Martin Liska <mliska@suse.cz>
12954 * passes.def: Include pass_lower_switch.
12955 * stmt.c (dump_case_nodes): Remove and move to
12956 tree-switch-conversion.
12957 (case_values_threshold): Likewise.
12958 (expand_switch_as_decision_tree_p): Likewise.
12959 (emit_case_decision_tree): Likewise.
12960 (expand_case): Likewise.
12961 (balance_case_nodes): Likewise.
12962 (node_has_low_bound): Likewise.
12963 (node_has_high_bound): Likewise.
12964 (node_is_bounded): Likewise.
12965 (emit_case_nodes): Likewise.
12966 (struct simple_case_node): New struct.
12967 (add_case_node): Remove.
12968 (emit_case_dispatch_table): Use vector instead of case_list.
12969 (reset_out_edges_aux): Remove.
12970 (compute_cases_per_edge): Likewise.
12971 (expand_case): Build list of simple_case_node.
12972 (expand_sjlj_dispatch_table): Use it.
12973 * tree-switch-conversion.c (struct case_node): Moved from
12974 stmt.c and adjusted.
12975 (emit_case_nodes): Likewise.
12976 (node_has_low_bound): Likewise.
12977 (node_has_high_bound): Likewise.
12978 (node_is_bounded): Likewise.
12979 (case_values_threshold): Likewise.
12980 (reset_out_edges_aux): Likewise.
12981 (compute_cases_per_edge): Likewise.
12982 (add_case_node): Likewise.
12983 (dump_case_nodes): Likewise.
12984 (balance_case_nodes): Likewise.
12985 (expand_switch_as_decision_tree_p): Likewise.
12986 (emit_jump): Likewise.
12987 (emit_case_decision_tree): Likewise.
12988 (try_switch_expansion): Likewise.
12989 (do_jump_if_equal): Likewise.
12990 (emit_cmp_and_jump_insns): Likewise.
12991 (fix_phi_operands_for_edge): New function.
12992 (record_phi_operand_mapping): Likewise.
12993 (class pass_lower_switch): New pass.
12994 (pass_lower_switch::execute): New function.
12995 (make_pass_lower_switch): Likewise.
12996 (conditional_probability):
12997 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12998 * tree-pass.h: Add make_pass_lower_switch.
13000 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13003 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13006 2017-08-29 Richard Biener <rguenther@suse.de>
13008 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13009 we do not add a DW_AT_inline attribute twice.
13010 (gen_subprogram_die): Remove code setting DW_AT_inline on
13011 DECL_ABSTRACT_P nodes.
13013 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13015 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13016 calls to internal functions.
13017 (gimplify_modify_expr): Likewise.
13018 * tree-call-cdce.c (use_internal_fn): Likewise.
13019 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13020 (convert_to_divmod): Set the nothrow flag.
13021 * tree-if-conv.c (predicate_mem_writes): Likewise.
13022 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13023 (vectorizable_call): Likewise.
13024 (vectorizable_store): Likewise.
13025 (vectorizable_load): Likewise.
13026 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13027 (vect_recog_mask_conversion_pattern): Likewise.
13029 2017-08-29 Martin Liska <mliska@suse.cz>
13032 * gcc.c (driver_handle_option): Add new argument.
13033 * opts-common.c (handle_option): Pass
13034 target_option_override_hook.
13035 * opts-global.c (lang_handle_option): Add new option.
13036 (set_default_handlers): Add new argument.
13037 (decode_options): Likewise.
13038 * opts.c (target_handle_option): Likewise.
13039 (common_handle_option): Call target_option_override_hook.
13040 * opts.h (struct cl_option_handler_func): Add hook for
13041 target option override.
13042 (struct cl_option_handlers): Likewise.
13043 (set_default_handlers): Add new argument.
13044 (decode_options): Likewise.
13045 (common_handle_option): Likewise.
13046 (target_handle_option): Likewise.
13047 * toplev.c (toplev::main): Pass targetm.target_option.override
13050 2017-08-29 Richard Biener <rguenther@suse.de>
13051 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13053 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13054 life to the active subtree.
13056 2017-08-28 Jeff Law <law@redhat.com>
13058 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13059 derive_equivalences.
13060 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13061 Code moved into....
13062 (edge_info::derive_equivalences): New private member function
13064 * tree-ssa-dom.c (class edge_info): Changed from a struct
13065 to a class. Add ctor/dtor, methods and data members.
13066 (edge_info::edge_info): Renamed from allocate_edge_info.
13067 Initialize additional members.
13068 (edge_info::~edge_info): New.
13069 (free_dom_edge_info): Delete the edge info.
13070 (record_edge_info): Use new class & associated member functions.
13071 Tighten forms for testing for edge equivalences.
13072 (record_temporary_equivalences): Iterate over the simple
13073 equivalences rather than assuming there's only one per edge.
13074 (cprop_into_successor_phis): Iterate over the simple
13075 equivalences rather than assuming there's only one per edge.
13076 (optimize_stmt): Use operand_equal_p rather than pointer
13077 equality for mini-DSE code.
13079 2017-08-28 Nathan Sidwell <nathan@acm.org>
13081 * gcc.c (execute): Fold SIGPIPE handling into switch
13082 statement. Adjust internal error message.
13084 2017-08-28 Richard Biener <rguenther@suse.de>
13087 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13088 Do nothing for removed DIEs.
13090 2017-08-28 Richard Biener <rguenther@suse.de>
13092 PR tree-optimization/81977
13093 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13096 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13099 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13100 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13101 using targetm.gen_mem_thread_fence.
13103 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13106 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13107 predicate to register_operand. Reorder operands.
13108 (*btr<mode>): Ditto.
13109 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13110 (*btr<mode>_mask): Ditto.
13112 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13114 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13115 * config/rs6000/xmmintrin.h: New file.
13116 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13118 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13121 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13122 parameter and_insn and return it.
13123 (recombine_lvx_pattern): Insert a copy to ensure availability of
13124 the base register of the copied masking operation at the point of
13125 the instruction replacement.
13126 (recombine_stvx_pattern): Likewise.
13128 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13130 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13131 undocumented switches.
13132 (-mpower9-dform-vector): Likewise.
13133 (-mpower9-dform): Likewise.
13134 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13135 comments to delete references to -mpower9-dform* switches.
13136 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13137 Delete reference to -mpower9-dform* switches, test for
13138 -mpower9-vector instead.
13139 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13140 (OTHER_P9_VECTOR_MASKS): Likewise.
13141 (POWERPC_MASKS): Likewise.
13142 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13143 tests against -mpower9-dform* to -mpower9-vector. Delete code
13144 that checked for -mpower9-dform* consistancy with other options.
13145 Add test for -mpower9-misc to enable other power9 switches.
13146 (rs6000_init_hard_regno_mode_ok): Likewise.
13147 (rs6000_option_override_internal): Likewise.
13148 (rs6000_emit_prologue): Likewise.
13149 (rs6000_emit_epilogue): Likewise.
13150 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13151 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13152 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13154 (emit_fusion_p9_store): Likewise.
13155 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13156 resetting these macros if the assembler does not support ISA 3.0
13158 (TARGET_P9_DFORM_VECTOR): Likewise.
13159 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13160 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13162 2017-08-25 Alan Modra <amodra@gmail.com>
13164 PR rtl-optimization/81747
13165 * cse.c (cse_extended_basic_block): Don't attempt to record
13166 equivalences for degenerate conditional branches that branch
13167 to their fall-through.
13169 2017-08-24 Martin Sebor <msebor@redhat.com>
13171 PR middle-end/81908
13172 * gimple-fold.c (size_must_be_zero_p): New function.
13173 (gimple_fold_builtin_memory_op): Call it.
13175 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13177 * config/rs6000/mm_malloc.h: New file.
13179 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13181 PR tree-optimization/81913
13182 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13183 analysis when either IVs in condition can wrap.
13185 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13187 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13188 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13190 2017-08-24 Richard Biener <rguenther@suse.de>
13193 * targhooks.c (default_target_can_inline_p): Properly
13194 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13195 is present and always compare.
13196 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13197 infer -mfpmath=sse from TARGET_SSE_P.
13198 (ix86_can_inline_p): Properly use target_option_default_node when
13199 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13201 2017-08-24 Richard Biener <rguenther@suse.de>
13204 * dwarf2out.c (output_die): Handle flag_generate_offload like
13206 (output_comp_unit): Likewise.
13207 (gen_array_type_die): Likewise.
13208 (dwarf2out_early_finish): Likewise.
13209 (note_variable_value_in_expr): Likewise.
13210 (dwarf2out_finish): Likewise. Adjust assert.
13211 * cgraphunit.c (symbol_table::compile): Move setting of
13212 flag_generate_offload earlier ...
13213 (symbol_table::finalize_compilation_unit): ... here, before
13214 early debug finalization.
13216 2017-08-24 Richard Biener <rguenther@suse.de>
13218 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13219 and ipa-fnsummary.h.
13220 (ix86_can_inline_p): When ix86_fpmath flags do not match
13221 check whether the callee uses FP math at all.
13223 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13225 PR middle-end/81931
13226 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13227 instead of TYPE_PRECISION.
13229 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13230 Alan Hayward <alan.hayward@arm.com>
13231 David Sherwood <david.sherwood@arm.com>
13233 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13235 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13237 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13239 (ix86_frame::stack_realign_allocate): New field.
13240 (struct machine_frame_state): Modify comments.
13241 (machine_frame_state::sp_realigned_fp_end): New field.
13242 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13243 layout calculation.
13244 (sp_valid_at): Add assertion to assure no attempt to access invalid
13245 offset of a realigned stack.
13246 (fp_valid_at): Likewise.
13247 (choose_baseaddr): Modify comments.
13248 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13249 ix86_expand_prologue.
13250 (ix86_expand_prologue): Modify stack realignment and allocation.
13251 (ix86_expand_epilogue): Modify comments.
13252 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13253 avx2_runtime, avx512f, and avx512f_runtime.
13255 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13257 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13258 (mstackrealign): Do not init to -1.
13259 * config/i386/i386.c (ix86_option_override_internal):
13260 Check opts_set, not opts when setting default value of
13261 opts->x_ix86_force_align_arg_pointer.
13263 2017-08-23 Richard Biener <rguenther@suse.de>
13265 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13266 lang_hooks.decl_printable_name.
13267 * print-rtl-function.c (print_rtx_function): Likewise.
13268 * tree-pretty-print.c (dump_function_header): Likewise.
13270 2017-08-23 Richard Biener <rguenther@suse.de>
13273 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13274 -g0 at compile-time.
13276 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13278 PR middle-end/19706
13279 * doc/sourcebuild.texi (Other hardware attributes):
13282 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13284 PR middle-end/19706
13285 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13286 Fix single-use check.
13288 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13290 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13292 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13294 * doc/install.texi: Modify to add more details on running selected
13297 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13299 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13300 is combined with -mabi=ms.
13301 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13304 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13306 PR tree-optimization/81488
13307 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13308 and cached_basis fields.
13309 (MAX_SPREAD): New constant.
13310 (alloc_cand_and_find_basis): Initialize new fields.
13311 (clear_visited): New function.
13312 (create_phi_basis_1): Rename from create_phi_basis, set visited
13313 and cached_basis fields.
13314 (create_phi_basis): New wrapper function.
13315 (phi_add_costs_1): Rename from phi_add_costs, add spread
13316 parameter, set visited field, short-circuit when limits reached.
13317 (phi_add_costs): New wrapper function.
13318 (record_phi_increments_1): Rename from record_phi_increments, set
13320 (record_phi_increments): New wrapper function.
13321 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13322 (phi_incr_cost): New wrapper function.
13323 (all_phi_incrs_profitable_1): Rename from
13324 all_phi_incrs_profitable, set visited field.
13325 (all_phi_incrs_profitable): New wrapper function.
13327 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13328 Alan Hayward <alan.hayward@arm.com>
13329 David Sherwood <david.sherwood@arm.com>
13331 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13332 that takes the outer and inner modes.
13333 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13334 comparison as the canonical test for a paradoxical subreg.
13335 * combine.c (simplify_set): Use paradoxical_subreg_p.
13336 (make_extraction): Likewise.
13337 (force_to_mode): Likewise.
13338 (rtx_equal_for_field_assignment_p): Likewise.
13339 (gen_lowpart_for_combine): Likewise.
13340 (simplify_comparison): Likewise.
13341 * cse.c (equiv_constant): Likewise.
13342 * expmed.c (store_bit_field_1): Likewise.
13343 * final.c (alter_subreg): Likewise.
13344 * fwprop.c (propagate_rtx): Likewise.
13345 (forward_propagate_subreg): Likewise.
13346 * ira-conflicts.c (ira_build_conflicts): Likewise.
13347 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13348 * lra-constraints.c (curr_insn_transform): Likewise.
13349 (split_reg): Likewise.
13350 * lra-eliminations.c (move_plus_up): Likewise.
13351 (lra_eliminate_regs_1): Likewise.
13352 * recog.c (general_operand): Likewise.
13353 * ree.c (combine_reaching_defs): Likewise.
13354 * reload.c (push_reload): Likewise.
13355 (find_reloads): Likewise.
13356 * reload1.c (elimination_effects): Likewise.
13357 (compute_reload_subreg_offset): Likewise.
13358 (choose_reload_regs): Likewise.
13359 * rtlanal.c (subreg_lsb_1): Likewise.
13360 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13361 (simplify_subreg): Likewise.
13362 * var-tracking.c (track_loc_p): Likewise.
13363 * emit-rtl.c (byte_lowpart_offset): Likewise.
13364 (paradoxical_subreg_p): Delete out-of-line definition.
13366 2017-08-22 Jeff Law <law@redhat.com>
13368 PR tree-optimization/81741
13369 PR tree-optimization/71947
13370 * tree-ssa-dom.c: Include tree-inline.h.
13371 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13372 equivalences if one is more expensive to compute than the other.
13373 * tree-ssa-scopedtables.h (class const_or_copies): Make
13374 record_const_or_copy_raw method private.
13375 (class avail_exprs_stack): New method simplify_binary_operation.
13376 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13377 avail_exprs_stack::simplify_binary_operation as needed.
13378 (avail_exprs_stack::simplify_binary_operation): New function.
13380 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13382 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13383 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13384 (DOT_SYMBOLS): Likewise.
13385 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13386 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13387 (RS6000_ABI_NAME): Likewise.
13388 (TARGET_CMODEL): Likewise.
13389 (TOC_SECTION_ASM_OP): Likewise.
13390 (SET_CMODEL): New macro.
13391 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13393 2017-08-22 Richard Biener <rguenther@suse.de>
13395 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13396 to allow for free-lang-data replacements similar to verify_type_variant.
13398 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13399 Alan Hayward <alan.hayward@arm.com>
13400 David Sherwood <david.sherwood@arm.com>
13402 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13403 VOIDmode for the LABEL_REF.
13405 2017-08-22 Richard Biener <rguenther@suse.de>
13407 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13409 2017-08-22 Richard Biener <rguenther@suse.de>
13411 * common.opt (feliminate-dwarf2-dups): Ignore.
13412 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13413 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13414 same_die_p_wrap, compute_section_prefix,
13415 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13416 (comdat_symbol_id, comdat_symbol_number): Likewise.
13417 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13419 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13420 (output_die): Mark unreachable path unreachable.
13421 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13422 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13423 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13424 (dwarf2out_early_finish): Likewise.
13426 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13428 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13430 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13433 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13434 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13435 (avr_attribute_table) <io, io_low, address>: Initialize
13436 .decl_required with true.
13438 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13440 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13441 undocumented debugging options.
13442 (-mvsx-scalar-double): Likewise.
13443 (-mallow-df-permute): Likewise.
13444 (-mvectorize-builtins): Likewise.
13445 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13446 (rs6000_builtin_vectorized_function): Likewise.
13447 (rs6000_builtin_md_vectorized_function): Likewise.
13448 (rs6000_opt_vars): Likewise.
13450 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13453 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13454 (*btrq_imm): Rename from *btrq.
13455 (*btcq_imm): Rename from *btcq.
13456 (btsc): New code attribute.
13457 (*<btsc><mode>): New insn pattern.
13458 (*btr<mode>): Ditto.
13459 (*<btsc><mode>_mask): New insn_and_split pattern.
13460 (*btr<mode>_mask): Ditto.
13462 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13463 Alan Hayward <alan.hayward@arm.com>
13464 David Sherwood <david.sherwood@arm.com>
13466 * function.c (pad_below): Simplify padding calculation.
13468 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13469 Alan Hayward <alan.hayward@arm.com>
13470 David Sherwood <david.sherwood@arm.com>
13472 * target.def (function_prologue): Remove frame size argument.
13473 (function_epilogue): Likewise.
13474 * doc/tm.texi: Regenerate.
13475 * final.c (final_start_function): Update call to function_prologue.
13476 (final_end_function): Update call to function_epilogue.
13477 (default_function_pro_epilogue): Remove frame size argument.
13478 * output.h (default_function_pro_epilogue): Likewise.
13479 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13480 (arm_output_function_prologue): Likewise.
13481 * config/frv/frv.c (frv_function_prologue): Likewise.
13482 (frv_function_epilogue): Likewise.
13483 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13484 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13485 (ia64_output_function_epilogue): Likewise.
13486 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13487 (m32r_output_function_epilogue): Likewise.
13488 * config/microblaze/microblaze.c (microblaze_function_prologue)
13489 (microblaze_function_epilogue): Likewise.
13490 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13491 (mips_output_function_epilogue): Likewise.
13492 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13493 (mmix_target_asm_function_epilogue): Likewise.
13494 * config/msp430/msp430.c (msp430_start_function): Likewise.
13495 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13496 (nds32_asm_function_epilogue): Likewise.
13497 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13498 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13499 (pa_output_function_epilogue): Likewise.
13500 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13501 (rs6000_output_function_epilogue): Likewise.
13502 * config/rl78/rl78.c (rl78_start_function): Likewise.
13503 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13504 (rs6000_output_function_epilogue): Likewise.
13505 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13506 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13507 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13508 (sparc_asm_function_epilogue): Likewise.
13510 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13512 * tree.h (type_has_mode_precision_p): New function.
13513 * convert.c (convert_to_integer_1): Use it.
13514 * expr.c (expand_expr_real_2): Likewise.
13515 (expand_expr_real_1): Likewise.
13516 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13517 * match.pd: Likewise.
13518 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13519 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13520 * tree-tailcall.c (process_assignment): Likewise.
13521 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13522 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13523 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13524 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13525 (vectorizable_assignment): Likewise.
13526 (vectorizable_shift): Likewise.
13527 (vectorizable_operation): Likewise.
13528 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13530 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13532 * match.pd: Add pow (C, x) simplification.
13534 2017-08-21 Richard Biener <rguenther@suse.de>
13536 PR tree-optimization/81900
13537 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13538 for blocks with abnormal predecessors.
13539 (compute_antic): Do not set visited flag prematurely.
13541 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13544 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13546 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13548 * stor-layout.h (vector_type_mode): Move to...
13549 * tree.h (vector_type_mode): ...here.
13550 * stor-layout.c (vector_type_mode): Move to...
13551 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13553 2017-08-21 Richard Biener <rguenther@suse.de>
13555 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
13556 register_external_die hooks.
13557 (debug_false_tree_charstarstar_uhwistar): Declare.
13558 (debug_nothing_tree_charstar_uhwi): Likewise.
13559 * debug.c (do_nothing_debug_hooks): Adjust.
13560 (debug_false_tree_charstarstar_uhwistar): New do nothing.
13561 (debug_nothing_tree_charstar_uhwi): Likewise.
13562 * dbxout.c (dbx_debug_hooks): Adjust.
13563 (xcoff_debug_hooks): Likewise.
13564 * sdbout.c (sdb_debug_hooks): Likewise.
13565 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13566 * dwarf2out.c (macinfo_label_base): New global.
13567 (dwarf2out_register_external_die): New function for the
13568 register_external_die hook.
13569 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
13570 (dwarf2_debug_hooks): Use them.
13571 (dwarf2_lineno_debug_hooks): Adjust.
13572 (struct die_struct): Add with_offset flag.
13573 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
13574 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
13575 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
13576 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
13577 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
13578 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
13579 defining section names for the early LTO debug variants.
13580 (reset_indirect_string): New helper.
13581 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
13582 (print_dw_val): Add support for offsetted symbol references.
13583 (get_ultimate_context): Split out from is_cxx.
13584 (is_cxx): Use get_ultimate_context.
13585 (is_fortran): Add decl overload.
13586 (compute_comp_unit_symbol): Split out worker from
13587 compute_section_prefix.
13588 (compute_section_prefix): Call compute_comp_unit_symbol and
13589 set comdat_type_p here.
13590 (output_die): Skip DIE symbol output for the LTO added one.
13591 Handle DIE symbol references with offset.
13592 (output_comp_unit): Guard section name mangling properly.
13593 For LTO debug sections emit a symbol at the section beginning
13594 which we use to refer to its DIEs.
13595 (add_abstract_origin_attribute): For DIEs registered via
13596 dwarf2out_register_external_die directly refer to the early
13597 DIE rather than indirectly through the shadow one we created.
13598 Remove obsolete call to dwarf2out_abstract_function for
13599 non-function/block origins.
13600 (gen_array_type_die): When generating early LTO debug do
13601 not emit DW_AT_string_length.
13602 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
13603 late when in LTO. As suggested place a gcc_unreachable for
13604 the DECL_ABSTRACT_P case.
13605 (gen_subprogram_die): Avoid another specification DIE
13606 for early built declarations/definitions for the late LTO case.
13607 (gen_variable_die): Add type references for late duplicated VLA dies
13609 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
13610 we have the abstract instance already.
13611 (process_scope_var): Adjust decl DIE contexts in LTO which
13612 first puts them in limbo.
13613 (gen_decl_die): Do not generate type DIEs late apart from
13614 types for VLAs or for decls we do not yet have a DIE. Do not
13615 call dwarf2out_abstract_function late.
13616 (dwarf2out_early_global_decl): Make sure to create DIEs
13617 for abstract instances of a decl first.
13618 (dwarf2out_late_global_decl): Adjust comment.
13619 (output_macinfo_op): With multiple macro sections use
13620 macinfo_label_base to distinguish labels.
13621 (output_macinfo): Likewise. Update macinfo_label_base.
13622 Pass in the line info label.
13623 (note_variable_value_in_expr): When generating LTO resolve
13624 all variable values here by generating DIEs as needed.
13625 (init_sections_and_labels): Add early LTO debug flag parameter
13626 and generate different sections and names if set. Add generation
13627 counter for the labels so we can have multiple of them.
13628 (reset_dies): Helper to allow DIEs to be output multiple times.
13629 (dwarf2out_finish): When outputting DIEs to the fat part of an
13630 LTO object first reset DIEs.
13631 (dwarf2out_early_finish): Output early DIEs when generating LTO.
13632 (modified_type_die): Check for decl_ultimate_origin being self
13634 (gen_type_die_with_usage): Likewise.
13635 (gen_typedef_die): Allow decl_ultimate_origin being self.
13636 (set_decl_abstract_flags): Remove.
13637 (set_block_abstract_flags): Likewise.
13638 (dwarf2out_abstract_function): Treat the early generated DIEs
13639 as the abstract copy and only add DW_AT_inline and
13640 DW_AT_artificial here and call set_decl_origin_self.
13641 If the DIE has an abstract origin don't do anything.
13642 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
13643 if we have none yet (Go fails to build one, PR78628).
13644 (variably_modified_type_p): Prevent endless recursion for Ada
13645 cyclic pointer types.
13646 * lto-streamer-in.c: Include debug.h.
13647 (dref_queue): New global.
13648 (lto_read_tree_1): Stream in DIE references.
13649 (lto_input_tree): Register DIE references.
13650 (input_function): Stream DECL_DEBUG_ARGS.
13651 * lto-streamer-out.c: Include debug.h.
13652 (lto_write_tree_1): Output DIE references.
13653 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
13654 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
13655 (output_function): Stream DECL_DEBUG_ARGS.
13656 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
13657 Stream DECL_ABSTRACT_ORIGIN.
13658 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
13659 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
13660 DECL_CONTEXT for file-scope decls.
13661 * lto-streamer.h (struct dref_entry): Declare.
13662 (dref_queue): Likewise.
13663 * cfgexpand.c (pass_expand::execute): Do not call the
13664 outlining_inline_function hook here.
13665 * lto-wrapper.c (debug_obj): New global.
13666 (tool_cleanup): Unlink it if required.
13667 (debug_objcopy): New function.
13668 (run_gcc): Handle early debug sections in the IL files by
13669 extracting them to separate files, partially linkin them and
13670 feeding the result back as result to the linker.
13671 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
13672 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
13673 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
13674 sections into a separate segment.
13675 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
13677 (darwin_asm_dwarf_section): Likewise.
13678 (darwin_asm_output_dwarf_offset): Likewise.
13679 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
13681 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13682 Alan Hayward <alan.hayward@arm.com>
13683 David Sherwood <david.sherwood@arm.com>
13685 * read-md.h (md_reader::record_potential_iterator_use): Replace
13686 pointer argument with an rtx and an index.
13687 * read-rtl.c (iterator_group::apply_iterator): Likewise.
13688 (apply_mode_iterator): Likewise.
13689 (apply_code_iterator): Likewise.
13690 (apply_int_iterator): Likewise.
13691 (apply_subst_iterator): Likewise.
13692 (record_iterator_use): Likewise.
13693 (record_attribute_use): Likewise.
13694 (md_reader::record_potential_iterator_use): Likewise. Update calls
13695 to record_iterator_use and apply_iterator.
13696 (iterator_use): Replace ptr with x and index.
13697 (attribute_use): Likewise.
13698 (apply_attribute_uses): Update calls to apply_iterator.
13699 (apply_iterators): Likewise. Update initialization of iterator_use.
13700 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
13701 and record_potential_iterator_use.
13702 (rtx_reader::read_rtx_operand): Likewise.
13704 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13705 Alan Hayward <alan.hayward@arm.com>
13706 David Sherwood <david.sherwood@arm.com>
13708 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
13711 2017-08-21 Richard Biener <rguenther@suse.de>
13713 PR middle-end/81884
13714 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
13715 at struct end conservatively when comparing common bases.
13717 2017-08-21 Richard Biener <rguenther@suse.de>
13719 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
13720 (mem_ref_in_stmt): Remove.
13721 (determine_max_movement): Use ref index to get at the reference.
13722 (invariantness_dom_walker::before_dom_children): Deal with
13723 lim data already initialized.
13724 (gather_mem_refs_stmt): Initialize lim data and record ref index.
13726 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13728 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
13729 (TARGET_ISA_ROUND): Ditto.
13730 (TARGET_ROUND): Ditto.
13731 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
13732 * config/i386/i386.md: Ditto.
13733 * config/i386/sse.md: Ditto.
13734 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
13735 with OPTION_MASK_ISA_SSE4_1.
13737 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13740 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13741 __builtin_ia32_lzcnt_u16.
13743 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13746 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13747 (rs6000_set_current_function): Rewrite function to use it.
13749 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13752 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13753 and TYPE_WARN_IF_NOT_ALIGN.
13754 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13755 (handle_warn_if_not_align): New.
13756 (place_union_field): Call handle_warn_if_not_align.
13757 (place_field): Call handle_warn_if_not_align.
13758 Copy TYPE_WARN_IF_NOT_ALIGN.
13759 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13760 (layout_type): Likewise.
13761 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13763 (tree_decl_common): Add warn_if_not_align.
13764 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13765 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13766 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13767 (DECL_WARN_IF_NOT_ALIGN): Likewise.
13768 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
13769 * doc/extend.texi: Document warn_if_not_aligned attribute.
13770 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
13772 2017-08-17 Martin Liska <mliska@suse.cz>
13775 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
13776 (get_data_dependence): Use it as pointer type.
13777 (distribute_loop): Likewise.
13779 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13781 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
13782 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
13783 (p8_vmrgow_<mode>_direct): New define_insn.
13784 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
13785 handle endianness for vmrgew and vmrgow permute patterns.
13787 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13789 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
13790 * config/rs6000/rs6000-cpus.def: Remove comment.
13791 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
13792 (POWERPC_MASKS): Likewise.
13793 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
13794 use of TARGET_VSX_TIMODE.
13795 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
13796 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
13797 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
13798 (rs6000_option_override_internal): Remove dead code.
13799 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
13800 (rs6000_legitimize_reload_address): Likewise.
13801 (rs6000_legitimate_address_p): Likewise.
13802 (rs6000_opt_masks): Delete "vsx-timode".
13803 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
13804 from function comment.
13805 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
13806 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
13807 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
13809 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
13810 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
13811 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
13812 (VSX_TI): Likewise.
13814 (define_peephole2): Likewise.
13816 2017-08-17 Martin Sebor <msebor@redhat.com>
13819 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13820 past the end of an array.
13821 (test_pp_format): Add test cases.
13823 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13825 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13826 missing ECF_NOTHROW flags.
13828 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13831 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13832 operands residing in integer registers.
13833 (*vsx_le_perm_load_<mode>): Likewise.
13834 (*vsx_le_perm_store_<mode>): Likewise.
13835 (define_peephole2): Add peepholes to optimize the above.
13837 2017-08-17 Marek Polacek <polacek@redhat.com>
13839 PR middle-end/81814
13840 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13841 to mimic what shorten_compare did. Change the return type to bool.
13842 (fold_cond_expr_with_comparison): Update call to
13843 operand_equal_for_comparison_p.
13844 (fold_ternary_loc): Likewise.
13846 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13848 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13850 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13852 2017-08-17 Richard Biener <rguenther@suse.de>
13854 * tree-ssa-structalias.c (solve_graph): When propagating
13855 to successors update the graphs succ edges and avoid duplicate work.
13857 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13860 * config/i386/i386.c (ix86_option_override_internal): Save target
13861 specific options after ix86_stack_protector_guard_reg was changed.
13863 2017-08-17 Richard Biener <rguenther@suse.de>
13865 PR tree-optimization/81827
13866 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13867 (new_var_info): Initialize it conservatively.
13868 (get_call_vi): Mark register vars.
13869 (new_scalar_tmp_constraint_exp): Likewise.
13870 (handle_rhs_call): Likewise.
13871 (handle_const_call): Likewise.
13872 (create_function_info_for): Likewise.
13873 (solve_constraints): Sort varinfos to separate register from
13874 non-register vars to pack points-to solution bitmaps during
13877 2017-08-17 Marek Polacek <polacek@redhat.com>
13879 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13881 2017-08-17 Richard Biener <rguenther@suse.de>
13883 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13884 to true when overflow is undefined and we saturated the result.
13886 2017-08-17 Alan Modra <amodra@gmail.com>
13889 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13890 Don't use store multiple if only one reg needs saving.
13891 (interesting_frame_related_regno): New function.
13892 (rs6000_frame_related): Don't emit frame info for regs that
13894 (rs6000_emit_epilogue): Likewise.
13896 2017-08-16 Nathan Sidwell <nathan@acm.org>
13898 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13899 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13900 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13901 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13902 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13903 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13904 (verify_type): Adjust for TYPE_BINFO move.
13905 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13906 process TYPE_BINFO directly.
13907 (hash_tree): Likewise.
13908 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13910 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13913 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13915 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13917 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13920 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13921 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13922 Add ix86_binary_operator_ok to insn constraint.
13923 (*iordi_1_bts): Ditto.
13924 (*xordi_1_btc): Ditto.
13925 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13926 Update corresponding peephole2 pattern.
13930 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13932 PR tree-optimization/81832
13933 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13934 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13936 2017-08-16 Marek Polacek <polacek@redhat.com>
13939 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13940 regarding size_zero_node.
13942 2017-08-16 Martin Liska <mliska@suse.cz>
13945 * config.gcc: Respect previously set extra_objs in case
13948 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13950 PR tree-optimization/81835
13951 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13952 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13953 not depend on the phi.
13955 2017-08-16 Alan Modra <amodra@gmail.com>
13957 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13960 2017-08-16 Alan Modra <amodra@gmail.com>
13962 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13963 (save_reg_p): ..into this. Update all callers.
13964 (first_reg_to_save): Simplify.
13966 2017-08-16 Alan Modra <amodra@gmail.com>
13968 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13971 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13975 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13976 constant count if that count is less than 32.
13978 2017-08-15 Nathan Sidwell <nathan@acm.org>
13980 * gcc.c (execute): Emit friendlier message if inferior is killed
13981 by an external cause.
13983 2017-08-15 Richard Biener <rguenther@suse.de>
13985 PR tree-optimization/81790
13986 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13987 CONSTRUCTORs from simplifying and VN.
13989 2017-08-14 Martin Sebor <msebor@redhat.com>
13991 * builtin-attrs.def: Add comments.
13993 2017-08-14 Martin Sebor <msebor@redhat.com>
13996 * doc/extend.texi (attribute nonstring): Document new attribute.
13998 2017-08-14 Martin Sebor <msebor@redhat.com>
14001 * tree-diagnostic.c (default_tree_printer): Handle %G.
14002 * gimple-pretty-print.h (percent_G_format): Declare new function.
14003 * gimple-pretty-print.c (percent_G_format): Define.
14004 * tree-pretty-print.c (percent_K_format): Add argument.
14006 2017-08-14 Martin Sebor <msebor@redhat.com>
14008 PR translation/79998
14009 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14010 Remove a stray space.
14012 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14015 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14016 (*iordi_1_bts): Ditto.
14017 (*xordi_1_btc): Ditto.
14019 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14022 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14023 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14025 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14026 quoted strings, and make more translator-friendly.
14027 (darwin_rs6000_override_options): Likewise.
14028 (rs6000_option_override_internal): Likewise.
14029 (rs6000_return_in_memory): Fix overlong line.
14030 (init_cmulative_args): Use quoted strings, and make more
14031 translator-friendly.
14032 (rs6000_pass_by_reference): Fix overlong line.
14033 (def_builtin): Use quoted strings.
14034 (altivec_expand_predicate_builtin): Use quoted strings, and make
14035 more translator-friendly.
14036 (htm_expand_builtin): Use quoted strings.
14037 (cpu_expand_builtin): Use quoted strings, and make more
14038 translator-friendly.
14039 (altivec_expand_builtin): Likewise.
14040 (paired_expand_predicate_builtin): Likewise.
14041 (rs6000_invalid_builtin): Likewise.
14042 (builtin_function_type): Use quoted strings.
14043 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14044 more translator-friendly.
14045 (rs6000_trampoline_init): Likewise.
14046 (rs6000_handle_altivec_attribute): Likewise.
14047 (rs6000_inner_target_options): Use quoted strings.
14048 (rs6000_disable_incompatible_switches): Likewise.
14049 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14050 strings, and make more translator-friendly.
14051 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14053 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14055 PR tree-optimization/81799
14056 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14057 cond_expr to simple gimple operand.
14059 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14061 PR middle-end/46932
14062 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14064 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14068 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14069 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14070 TARGET_GASISR_PROLOGUES.
14071 * config/avr/avr.c (avr_option_override): Same.
14072 (avr_pass_pre_proep::execute): Same.
14074 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14077 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14078 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14080 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14082 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14083 define_insn_and_split. Split to a memory load from 0 in
14084 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14085 using PTR mode iterator.
14086 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14087 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14088 (*add_tp_<mode>): Redefine as define_insn_and_split.
14089 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14090 address space. Merge with *add_tp_x32 using PTR mode iterator.
14091 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14092 Split to an add with a memory load from 0 in
14093 DEFAULT_TLS_SEG_REG address space.
14095 2017-08-12 Andrew Pinski <apinski@cavium.com>
14097 * config/aarch64/aarch64-option-extensions.def (rdma):
14098 Fix feature string to what Linux prints out in /proc/cpuinfo.
14100 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14103 * dwarf2out.c (modified_type_die): For C typedef types that have
14104 an ultimate origin, process the ultimate origin instead of the
14106 (gen_typedef_die): Assert that input DECLs have no ultimate
14108 (gen_type_die_with_usage): For typedef variants that have an
14109 ultimate origin, just call gen_decl_die on the original DECL.
14110 (process_scope_var): Avoid creating DIEs for local typedefs and
14111 concrete static variables.
14113 2017-08-12 Alan Modra <amodra@gmail.com>
14117 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14118 match gnu-user.h startfile.
14119 (ENDFILE_LINUX_SPEC): Similarly.
14121 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14124 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14126 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14128 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14129 * config/aarch64/aarch64.md (mov<mode>): Change.
14130 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14131 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14132 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14134 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14136 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14137 for storage order barriers.
14139 2017-08-11 Martin Liska <mliska@suse.cz>
14142 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14143 variables of void type.
14145 2017-08-11 Martin Liska <mliska@suse.cz>
14147 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14148 TARGET_SUPPORTS_ALIASES.
14149 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14150 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14151 (optimize_weakref): Likewise.
14152 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14153 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14154 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14156 2017-08-11 Martin Liska <mliska@suse.cz>
14159 * config/i386/i386.c (make_resolver_func): Do complete
14160 refactoring of the function.
14162 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14165 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14166 * config/i386/i386.c (ix86_stack_protect_guard): Use
14167 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14168 * doc/invoke.texi (x86 Options): Document
14169 -mstack-protector-guard-symbol= option.
14171 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14173 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14174 * config/i386/i386.c (ix86_split_stack_guard): New function.
14175 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14176 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14177 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14178 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14179 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14180 (split_stack_space_check): Call ix86_split_stack_guard.
14182 2017-08-10 Martin Sebor <msebor@redhat.com>
14184 * print-tree.c (print_node): Print location using the established
14186 Replace spaces with colons.
14187 (debug_raw, debug): Ditto.
14189 2017-08-10 Martin Sebor <msebor@redhat.com>
14192 * pretty-print.c (pp_format): Correct the handling of %s precision.
14194 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14197 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14199 (ix86_finalize_stack_frame_flags): This. Also clear
14200 frame_pointer_needed if -fno-omit-frame-pointer is used without
14202 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14203 with ix86_finalize_stack_frame_flags.
14204 (ix86_expand_epilogue): Likewise.
14205 (ix86_expand_split_stack_prologue): Likewise.
14206 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14208 2017-08-10 Martin Liska <mliska@suse.cz>
14211 * c-attribs.c (handle_target_attribute):
14212 Report warning for an empty string argument of target attribute.
14214 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14217 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14219 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14220 or DECL_NONLOCAL labels.
14221 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14222 or DECL_NONLOCAL labels here.
14224 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14226 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14227 to indicate when early gimple folding has been disabled.
14228 (rs6000_gimple_fold_builtin): Add debug content.
14229 (rs6000_invalid_builtin): Fix whitespace.
14230 (rs6000_expand_builtin): Fix whitespace.
14231 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14233 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14236 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14237 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14239 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14241 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14242 (qdf24xx): Likewise.
14243 * config/aarch64/aarch64.md: Include falkor.md.
14244 * config/aarch64/falkor.md: New.
14246 2017-08-09 Marek Polacek <polacek@redhat.com>
14249 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14250 * diagnostic.c (emit_diagnostic): Add a comment.
14251 (emit_diagnostic_valist): New function.
14253 2017-08-09 Marek Polacek <polacek@redhat.com>
14256 * input.c (make_location): New overload.
14257 * input.h (make_location): Declare.
14259 2017-08-08 Alan Modra <amodra@gmail.com>
14260 H.J. Lu <hongjiu.lu@intel.com>
14263 * gcc.c (NO_PIE_SPEC): Delete.
14264 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14266 (LINK_PIE_SPEC): ..to here.
14267 (LINK_COMMAND_SPEC): Support -no-pie.
14268 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14269 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14271 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14272 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14273 (ENDFILE_CRTEND_SPEC): Similarly.
14275 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14278 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14279 (mstack-protector-guard-offset=): Ditto.
14280 * config/i386/i386.c (ix86_option_override): Handle
14281 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14283 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14284 ix86_stack_protect_guard_offset variables.
14285 (TARGET_STACK_PROTECT_GUARD): Always define.
14286 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14287 and -mstack-protector-guard-offset= options.
14289 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14291 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14292 boundary case for the last candidate.
14294 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14296 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14297 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14299 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14301 PR middle-end/19706
14302 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14303 * config/aarch64/aarch64-builtins.c
14304 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14305 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14306 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14308 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14309 Andrew Pinski <pinskia@gmail.com>
14311 PR middle-end/19706
14312 * internal-fn.def (XORSIGN): New.
14313 * optabs.def (xorsign_optab): New.
14314 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14315 (convert_expand_mult_copysign): New.
14316 (pass_optimize_widening_mul::execute): Call
14317 convert_expand_mult_copysign.
14319 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14321 PR tree-optimization/81354
14322 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14323 Insert on edges rather than explicitly creating landing pads.
14324 (analyze_candidates_and_replace): Commit edge inserts.
14326 2017-08-08 Richard Biener <rguenther@suse.de>
14328 PR middle-end/81719
14329 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14330 (expand_simple_operations): Also look through ADDR_EXPRs with
14331 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14333 2017-08-08 Richard Biener <rguenther@suse.de>
14335 PR tree-optimization/81723
14336 * tree-vect-slp.c (struct bst_traits): New hash traits.
14337 (bst_fail): New global.
14338 (vect_build_slp_tree_2): New worker, split out from ...
14339 (vect_build_slp_tree): ... this now wrapping it with using
14340 bst_fail set to cache SLP tree build fails. Properly handle
14342 (vect_analyze_slp_instance): Allocate and free bst_fail.
14344 2017-08-08 Martin Liska <mliska@suse.cz>
14347 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14348 LABEL_DECLs that can be from a different function.
14350 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14352 PR tree-optimization/81744
14353 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14354 loop's number of iterations.
14356 2017-08-08 Martin Liska <mliska@suse.cz>
14358 * asan.c: Include header files.
14359 * attribs.c (build_decl_attribute_variant): New function moved
14361 (build_type_attribute_qual_variant): Likewise.
14362 (cmp_attrib_identifiers): Likewise.
14363 (simple_cst_list_equal): Likewise.
14364 (omp_declare_simd_clauses_equal): Likewise.
14365 (attribute_value_equal): Likewise.
14366 (comp_type_attributes): Likewise.
14367 (build_type_attribute_variant): Likewise.
14368 (lookup_ident_attribute): Likewise.
14369 (remove_attribute): Likewise.
14370 (merge_attributes): Likewise.
14371 (merge_type_attributes): Likewise.
14372 (merge_decl_attributes): Likewise.
14373 (merge_dllimport_decl_attributes): Likewise.
14374 (handle_dll_attribute): Likewise.
14375 (attribute_list_equal): Likewise.
14376 (attribute_list_contained): Likewise.
14377 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14378 (lookup_attribute_by_prefix): Likewise.
14379 * bb-reorder.c: Include header files.
14380 * builtins.c: Likewise.
14381 * calls.c: Likewise.
14382 * cfgexpand.c: Likewise.
14383 * cgraph.c: Likewise.
14384 * cgraphunit.c: Likewise.
14385 * convert.c: Likewise.
14386 * dwarf2out.c: Likewise.
14387 * final.c: Likewise.
14388 * fold-const.c: Likewise.
14389 * function.c: Likewise.
14390 * gimple-expr.c: Likewise.
14391 * gimple-fold.c: Likewise.
14392 * gimple-pretty-print.c: Likewise.
14393 * gimple.c: Likewise.
14394 * gimplify.c: Likewise.
14395 * hsa-common.c: Likewise.
14396 * hsa-gen.c: Likewise.
14397 * internal-fn.c: Likewise.
14398 * ipa-chkp.c: Likewise.
14399 * ipa-cp.c: Likewise.
14400 * ipa-devirt.c: Likewise.
14401 * ipa-fnsummary.c: Likewise.
14402 * ipa-inline.c: Likewise.
14403 * ipa-visibility.c: Likewise.
14405 * lto-cgraph.c: Likewise.
14406 * omp-expand.c: Likewise.
14407 * omp-general.c: Likewise.
14408 * omp-low.c: Likewise.
14409 * omp-offload.c: Likewise.
14410 * omp-simd-clone.c: Likewise.
14411 * opts-global.c: Likewise.
14412 * passes.c: Likewise.
14413 * predict.c: Likewise.
14414 * sancov.c: Likewise.
14415 * sanopt.c: Likewise.
14416 * symtab.c: Likewise.
14417 * toplev.c: Likewise.
14418 * trans-mem.c: Likewise.
14419 * tree-chkp.c: Likewise.
14420 * tree-eh.c: Likewise.
14421 * tree-into-ssa.c: Likewise.
14422 * tree-object-size.c: Likewise.
14423 * tree-parloops.c: Likewise.
14424 * tree-profile.c: Likewise.
14425 * tree-ssa-ccp.c: Likewise.
14426 * tree-ssa-live.c: Likewise.
14427 * tree-ssa-loop.c: Likewise.
14428 * tree-ssa-sccvn.c: Likewise.
14429 * tree-ssa-structalias.c: Likewise.
14430 * tree-ssa.c: Likewise.
14431 * tree-streamer-in.c: Likewise.
14432 * tree-vectorizer.c: Likewise.
14433 * tree-vrp.c: Likewise.
14434 * tsan.c: Likewise.
14435 * ubsan.c: Likewise.
14436 * varasm.c: Likewise.
14437 * varpool.c: Likewise.
14438 * tree.c: Remove functions moved to attribs.[ch].
14439 * tree.h: Likewise.
14440 * config/aarch64/aarch64.c: Add attrs.h header file.
14441 * config/alpha/alpha.c: Likewise.
14442 * config/arc/arc.c: Likewise.
14443 * config/arm/arm.c: Likewise.
14444 * config/avr/avr.c: Likewise.
14445 * config/bfin/bfin.c: Likewise.
14446 * config/c6x/c6x.c: Likewise.
14447 * config/cr16/cr16.c: Likewise.
14448 * config/cris/cris.c: Likewise.
14449 * config/darwin.c: Likewise.
14450 * config/epiphany/epiphany.c: Likewise.
14451 * config/fr30/fr30.c: Likewise.
14452 * config/frv/frv.c: Likewise.
14453 * config/ft32/ft32.c: Likewise.
14454 * config/h8300/h8300.c: Likewise.
14455 * config/i386/winnt.c: Likewise.
14456 * config/ia64/ia64.c: Likewise.
14457 * config/iq2000/iq2000.c: Likewise.
14458 * config/lm32/lm32.c: Likewise.
14459 * config/m32c/m32c.c: Likewise.
14460 * config/m32r/m32r.c: Likewise.
14461 * config/m68k/m68k.c: Likewise.
14462 * config/mcore/mcore.c: Likewise.
14463 * config/microblaze/microblaze.c: Likewise.
14464 * config/mips/mips.c: Likewise.
14465 * config/mmix/mmix.c: Likewise.
14466 * config/mn10300/mn10300.c: Likewise.
14467 * config/moxie/moxie.c: Likewise.
14468 * config/msp430/msp430.c: Likewise.
14469 * config/nds32/nds32-isr.c: Likewise.
14470 * config/nds32/nds32.c: Likewise.
14471 * config/nios2/nios2.c: Likewise.
14472 * config/nvptx/nvptx.c: Likewise.
14473 * config/pa/pa.c: Likewise.
14474 * config/pdp11/pdp11.c: Likewise.
14475 * config/powerpcspe/powerpcspe.c: Likewise.
14476 * config/riscv/riscv.c: Likewise.
14477 * config/rl78/rl78.c: Likewise.
14478 * config/rx/rx.c: Likewise.
14479 * config/s390/s390.c: Likewise.
14480 * config/sh/sh.c: Likewise.
14481 * config/sol2.c: Likewise.
14482 * config/sparc/sparc.c: Likewise.
14483 * config/spu/spu.c: Likewise.
14484 * config/stormy16/stormy16.c: Likewise.
14485 * config/tilegx/tilegx.c: Likewise.
14486 * config/tilepro/tilepro.c: Likewise.
14487 * config/v850/v850.c: Likewise.
14488 * config/vax/vax.c: Likewise.
14489 * config/visium/visium.c: Likewise.
14490 * config/xtensa/xtensa.c: Likewise.
14492 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14495 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14496 constraints since the -mupper-regs-* switches have been
14498 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14499 into a vector from a double word element that was extracted from
14500 another vector, and eliminate extra XXPERMDI instructions.
14501 (vsx_concat_<mode>_2): Likewise.
14502 (vsx_concat_<mode>_3): Likewise.
14503 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14504 concat to allow optimizing inserts from previous extracts.
14506 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14508 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14509 memory reference to a SSP offset in TLS address space.
14510 (ix86_print_operand) <case '@'>: Remove.
14511 (ix86_print_operand_punct_valid_p): Remove '@' code.
14512 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14513 UNSPEC_SP_TLS_TEST.
14514 (stack_tls_protect_set_<mode>): Remove.
14515 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14516 (stack_tls_protect_test_<mode>): Remove.
14517 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14519 2017-08-07 Olivier Hainque <hainque@adacore.com>
14522 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14524 2017-08-07 Douglas Rupp <rupp@adacore.com>
14526 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14527 variable was referenced as multidir in command.
14529 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14532 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14535 2017-08-07 Martin Liska <mliska@suse.cz>
14537 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14538 * config/rl78/rl78.c: Add include of attribs.h.
14539 * config/sh/sh.c: Likewise.
14540 * config/v850/v850.c: Likewise.
14542 2017-08-07 Tom de Vries <tom@codesourcery.com>
14544 PR middle-end/78266
14545 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14547 2017-08-07 Martin Liska <mliska@suse.cz>
14549 * config/mips/mips.c: Include attribs.h.
14551 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14554 * doc/invoke.texi: Document change in behvaior for -Ofast for
14557 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
14559 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
14561 (aarch64_pop_regs): Likewise.
14562 (aarch64_gen_load_pair): Likewise.
14563 (aarch64_save_callee_saves): Likewise.
14564 (aarch64_restore_callee_saves): Likewise.
14566 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14568 * config/i386/i386.c: Revert the last change.
14570 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14573 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14575 (ix86_finalize_stack_frame_flags): This. Also clear
14576 frame_pointer_needed if -fno-omit-frame-pointer is used without
14578 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14579 with ix86_finalize_stack_frame_flags.
14580 (ix86_expand_epilogue): Likewise.
14581 (ix86_expand_split_stack_prologue): Likewise.
14583 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14586 * config/i386/i386.c (get_builtin_code_for_version): Set priority
14587 to P_AES for Westmere.
14589 2017-08-07 Jonathan Yong <10walls@gmail.com>
14591 * config/i386/mingw.opt (fset-stack-executable): Removed.
14592 * config/i386/cygming.opt (fset-stack-executable): Moved
14594 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
14596 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
14598 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
14600 2017-08-07 Marek Polacek <polacek@redhat.com>
14602 PR middle-end/81737
14603 * fold-const.c (fold_indirect_ref_1): Check type_domain.
14605 2017-08-07 Martin Liska <mliska@suse.cz>
14607 * attribs.h (canonicalize_attr_name): New function.
14608 (cmp_attribs): Move from c-format.c and adjusted.
14609 (is_attribute_p): Moved from tree.h.
14610 * tree-inline.c: Add new includes.
14611 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
14612 (private_is_attribute_p): Remove.
14613 (private_lookup_attribute): Likewise.
14614 (private_lookup_attribute_by_prefix): Simplify.
14615 (remove_attribute): Use is_attribute_p.
14616 * tree.h: Remove removed declarations.
14618 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14620 PR middle-end/81698
14621 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
14622 instead of computing it in the function. Formatting fix.
14623 (expand_case): Don't rely on default_edge being the first edge,
14624 clear it if removing it, pass default_edge to
14625 emit_case_dispatch_table.
14626 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
14629 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
14631 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
14632 insn in the function, emit NOP after the insn.
14634 2017-08-06 Tom de Vries <tom@codesourcery.com>
14636 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
14639 2017-08-06 Tom de Vries <tom@codesourcery.com>
14641 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
14644 2017-08-04 Yury Gribov <tetra2005@gmail.com>
14646 PR tree-optimization/57371
14647 * match.pd: New pattern.
14649 2017-08-04 Marek Polacek <polacek@redhat.com>
14651 PR middle-end/81695
14652 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
14653 perform the computation in offset_int.
14655 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14657 PR tree-optimization/81136
14658 * tree-vectorizer.h: Include tree-hash-traits.h.
14659 (vec_base_alignments): New typedef.
14660 (vec_info): Add a base_alignments field.
14661 (vect_record_base_alignments): Declare.
14662 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
14664 (DR_IS_CONDITIONAL_IN_STMT): New macro.
14665 (create_data_ref): Add an is_conditional_in_stmt argument.
14666 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
14667 the is_conditional_in_stmt field.
14668 (data_ref_loc): Add an is_conditional_in_stmt field.
14669 (get_references_in_stmt): Set the is_conditional_in_stmt field.
14670 (find_data_references_in_stmt): Update call to create_data_ref.
14671 (graphite_find_data_references_in_stmt): Likewise.
14672 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
14673 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
14674 (vect_record_base_alignment): New function.
14675 (vect_record_base_alignments): Likewise.
14676 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
14677 for nested statements even if we fail to compute a misalignment.
14678 Use pooled base alignments for unconditional references.
14679 (vect_find_same_alignment_drs): Compare base addresses instead
14681 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
14682 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
14684 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14686 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
14687 Add an explicit name for the enum. Use auto_vec for slp_instances
14688 and grouped_stores.
14689 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
14691 (_bb_vec_info): Add a constructor and destructor.
14692 (vinfo_for_stmt): Return NULL for uids of -1 as well.
14693 (destroy_loop_vec_info): Delete.
14694 (vect_destroy_datarefs): Likewise.
14695 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
14696 (vec_info::vec_info): New function.
14697 (vec_info::~vec_info): Likewise.
14698 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
14699 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
14700 destroy_loop_vec_info.
14701 * tree-vect-loop.c (new_loop_vec_info): Replace with...
14702 (_loop_vec_info::_loop_vec_info): ...this.
14703 (destroy_loop_vec_info): Replace with...
14704 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
14705 the stmt_vec_infos. Leave handling of vec_info information to its
14706 destructor. Remove explicit vector releases.
14707 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
14708 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
14709 * tree-vect-slp.c (new_bb_vec_info): Replace with...
14710 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
14711 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
14712 (destroy_bb_vec_info): Replace with...
14713 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
14714 information to its destructor.
14715 (vect_slp_analyze_bb_1): Use new and delete instead of
14716 new_bb_vec_info and destroy_bb_vec_info.
14717 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
14720 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14722 * tree-data-ref.h (subscript): Add access_fn field.
14723 (data_dependence_relation): Add could_be_independent_p.
14724 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
14725 (same_access_functions): Move to tree-data-ref.c.
14726 * tree-data-ref.c (ref_contains_union_access_p): New function.
14727 (access_fn_component_p): Likewise.
14728 (access_fn_components_comparable_p): Likewise.
14729 (dr_analyze_indices): Add a reference to access_fn_component_p.
14730 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
14732 (constant_access_functions): Likewise.
14733 (add_other_self_distances): Likewise.
14734 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
14735 (initialize_data_dependence_relation): Use XCNEW and remove
14736 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
14737 of access functions that have the same type. Allow the
14738 subsequence to end with different bases in some circumstances.
14739 Record the chosen access functions in SUB_ACCESS_FN.
14740 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14741 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14742 (subscript_dependence_tester_1): Likewise dra and drb.
14743 (build_classic_dist_vector): Update calls accordingly.
14744 (subscript_dependence_tester): Likewise.
14745 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14746 DDR_COULD_BE_INDEPENDENT_P.
14747 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14748 comp_alias_ddrs instead of may_alias_ddrs.
14749 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14751 (vect_analyze_data_ref_dependence): Use it if
14752 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14753 distance vectors if that fails.
14754 (dependence_distance_ge_vf): New function.
14755 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14756 LOOP_VINFO_MAY_ALIAS_DDRS.
14758 2017-08-04 Richard Biener <rguenther@suse.de>
14760 PR middle-end/81705
14761 * fold-const.c (fold_binary_loc): Properly restrict
14762 minus_var0 && minus_var1 case when associating undefined overflow
14765 2017-08-04 Tom de Vries <tom@codesourcery.com>
14767 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
14769 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14771 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14772 Don't start diagnostic messages with a capital letter.
14773 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14775 (rs6000_invalid_builtin): Likewise.
14776 (rs6000_trampoline_init): Likewise.
14778 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14781 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
14782 after setting changeable df flags.
14784 2017-08-03 Richard Biener <rguenther@suse.de>
14786 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
14787 up if the use is in USE - X.
14789 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
14791 * toplev.c (dumpfile.h): New include.
14792 (internal_error_reentered): New static function. Use it...
14793 (internal_error_function): ...here to handle reentered internal_error.
14795 2017-08-03 Richard Biener <rguenther@suse.de>
14797 PR middle-end/81148
14798 * fold-const.c (split_tree): Add minus_var and minus_con
14799 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
14800 here but always use minus_*.
14801 (associate_trees): Assert we never associate with MINUS_EXPR
14802 and NULL first operand. Do not recurse for PLUS_EXPR operands
14803 when associating as MINUS_EXPR either.
14804 (fold_binary_loc): Track minus_var and minus_con.
14806 2017-08-03 Tom de Vries <tom@codesourcery.com>
14809 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
14810 ACCEL_COMPILER, apply finish_options on
14811 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
14813 2017-08-03 Tom de Vries <tom@codesourcery.com>
14816 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
14817 function_entry_patch_area_size > 0.
14819 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14822 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14823 instead of 10??LU, perform unit multiplication in wide_int,
14824 don't change alloc_object_size_limit if the limit is larger
14827 PR tree-optimization/81655
14828 PR tree-optimization/81588
14829 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14830 the case when ranges[i].low and high are 1 for unsigned type with
14833 PR middle-end/81052
14834 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14835 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14837 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14839 * tree-vrp.h: Add include guard.
14841 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14844 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14845 (ud2): New insn pattern.
14846 * config/i386/i386.c (ix86_expand_epilogue):
14847 For naked functions, generate ud2 instead of trap insn.
14849 2017-08-02 Marek Polacek <polacek@redhat.com>
14852 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14854 2017-08-02 Tom de Vries <tom@codesourcery.com>
14855 Cesar Philippidis <cesar@codesourcery.com>
14857 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14858 Add missing edge probabilities.
14860 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14862 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14863 Correct endianness.
14865 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14867 PR middle-end/79499
14868 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14869 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14870 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14871 prologue_seq sequences - if any.
14873 2017-08-02 Richard Biener <rguenther@suse.de>
14875 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14876 via vectors if supported, integer extracts via punning if supported
14877 or otherwise vector extracts.
14879 2017-08-02 Richard Biener <rguenther@suse.de>
14881 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14883 (bitmap_insert_into_set): ... this.
14885 2017-08-02 Richard Biener <rguenther@suse.de>
14887 PR tree-optimization/81633
14889 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14891 PR tree-optimization/71752
14892 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14894 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14896 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14897 (machine_function::call_ms2sysv_pad_out): Remove field.
14898 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14899 (ix86_compute_frame_layout): Likewise.
14901 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14904 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14905 attribute with interrupt attribute.
14907 2017-08-01 Andrew Pinski <apinski@cavium.com>
14909 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14910 BIT_INSERT_EXPR's operand 1
14911 to see if the types precision matches.
14913 2017-08-01 Martin Liska <mliska@suse.cz>
14915 PR middle-end/70140
14916 * builtins.c (expand_builtin_memcpy_args): Remove.
14917 (expand_builtin_memcpy): Call newly added function
14918 expand_builtin_memory_copy_args.
14919 (expand_builtin_memcpy_with_bounds): Likewise.
14920 (expand_builtin_mempcpy): Remove last argument.
14921 (expand_builtin_mempcpy_with_bounds): Likewise.
14922 (expand_builtin_memory_copy_args): New function created from
14923 expand_builtin_mempcpy_args with small modifications.
14924 (expand_builtin_mempcpy_args): Remove.
14925 (expand_builtin_stpcpy): Remove unused argument.
14926 (expand_builtin): Likewise.
14927 (expand_builtin_with_bounds): Likewise.
14929 2017-08-01 Martin Liska <mliska@suse.cz>
14932 Make mempcpy more optimal (PR middle-end/70140).
14934 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14937 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14938 __builtin_vec_cmpne verify both arguments are compatible vectors
14939 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14940 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14941 move computation of aligned to after checking the argument types.
14945 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14946 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14949 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14950 Doug Rupp <rupp@adacore.com>
14951 Olivier Hainque <hainque@adacore.com>
14953 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14954 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14956 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14957 for TARGET_OS_CPP_BUILTIN.
14958 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14959 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14961 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14962 passing required abi options to the assembler for EABI configurations.
14963 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14964 of .text.hot and .text.unlikely sections for kernel modules when
14965 using ARM style exceptions.
14966 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14967 options. Add EXTRA_CC1_SPEC.
14968 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14970 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14972 (ARM_TARGET2_DWARF_FORMAT): Define.
14973 * config/arm/t-vxworks: Adjust multilib control to removal of the
14974 Diab command line options.
14976 2017-08-01 Martin Liska <mliska@suse.cz>
14978 PR gcov-profile/81561
14979 * gcov.c (unblock): Make unblocking safe as we need to preserve
14980 index correspondence of blocks and block_lists.
14982 2017-08-01 Richard Biener <rguenther@suse.de>
14984 PR tree-optimization/81181
14985 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14986 (compute_antic): ... end of iteration here.
14988 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14990 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14991 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14992 (ftree-slp-vectorize): Likewise.
14993 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14994 can no longer be set independent of flag_tree_loop_vectorize.
14995 * omp-general.c (emp_max_vf): Likewise.
14996 * opts.c (enable_fdo_optimizations): Remove references to
14997 flag_tree_vectorize, these are now implicit.
14998 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14999 and leave it for the options machinery.
15001 2017-08-01 Martin Liska <mliska@suse.cz>
15003 PR middle-end/70140
15004 * builtins.c (expand_builtin_memcpy_args): Remove.
15005 (expand_builtin_memcpy): Call newly added function
15006 expand_builtin_memory_copy_args.
15007 (expand_builtin_memcpy_with_bounds): Likewise.
15008 (expand_builtin_mempcpy): Remove last argument.
15009 (expand_builtin_mempcpy_with_bounds): Likewise.
15010 (expand_builtin_memory_copy_args): New function created from
15011 expand_builtin_mempcpy_args with small modifications.
15012 (expand_builtin_mempcpy_args): Remove.
15013 (expand_builtin_stpcpy): Remove unused argument.
15014 (expand_builtin): Likewise.
15015 (expand_builtin_with_bounds): Likewise.
15017 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15020 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15021 print "ds:" only for immediates in generic address space.
15023 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15026 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15027 (ix86_function_ok_for_sibcall): Return false for naked functions.
15029 2017-08-01 Richard Biener <rguenther@suse.de>
15031 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15032 (compute_antic): Seed worklist with exit block predecessors.
15033 * cfganal.c (dfs_find_deadend): For a cycle return the source
15034 of the edge closing it.
15036 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15038 * config/aarch64/aarch64.c
15039 (aarch64_can_const_movi_rtx_p): Move 0 check.
15041 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15043 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15044 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15046 * match.pd: Ditto in address comparison pattern.
15048 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15050 PR tree-optimization/81627
15051 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15052 closed ssa form for store-store chain.
15054 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15056 PR tree-optimization/81620
15057 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15058 for store-store chain.
15060 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15062 PR tree-optimization/81588
15063 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15064 ranges[i].in_p, invert comparison code ccode. For >/>=,
15065 swap rhs1 and rhs2 and comparison code unconditionally,
15066 for </<= don't do that. Don't swap rhs1/rhs2 again if
15067 ranges[i].in_p, instead invert comparison code ccode if
15068 opcode or oe->rank is BIT_IOR_EXPR.
15071 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15072 a direct optab to conversion optab.
15073 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15074 with GET_MODE_INNER as last argument instead of optab_handler.
15075 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15076 vector extraction if possible and optab is available.
15077 * expr.c (store_constructor): Use convert_optab_handler instead
15078 of optab_handler. Use vector initialization from smaller
15079 vectors if possible and optab is available.
15080 * tree-vect-stmts.c (vectorizable_load): Likewise.
15081 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15082 now have two modes.
15083 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15084 of vec_init from half-sized vectors with the same element mode.
15085 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15086 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15087 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15088 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15089 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15090 after mode in gen_vec_extract* calls.
15091 (vec_extract<mode>): Renamed to ...
15092 (vec_extract<mode><ssescalarmodelower>): ... this.
15093 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15094 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15095 element mode after mode in gen_vec_init* calls.
15096 (VEC_INIT_HALF_MODE): New mode iterator.
15097 (vec_init<mode>): Renamed to ...
15098 (vec_init<mode><ssescalarmodelower>): ... this.
15099 (vec_init<mode><ssehalfvecmodelower>): New expander.
15100 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15101 (vec_extractv2sfsf): ... this.
15102 (vec_initv2sf): Renamed to ...
15103 (vec_initv2sfsf): ... this.
15104 (vec_extractv2si): Renamed to ...
15105 (vec_extractv2sisi): ... this.
15106 (vec_initv2si): Renamed to ...
15107 (vec_initv2sisi): ... this.
15108 (vec_extractv4hi): Renamed to ...
15109 (vec_extractv4hihi): ... this.
15110 (vec_initv4hi): Renamed to ...
15111 (vec_initv4hihi): ... this.
15112 (vec_extractv8qi): Renamed to ...
15113 (vec_extractv8qiqi): ... this.
15114 (vec_initv8qi): Renamed to ...
15115 (vec_initv8qiqi): ... this.
15116 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15117 (vec_init<mode>): Renamed to ...
15118 (vec_init<mode><VEC_base_l>): ... this.
15119 (vec_extract<mode>): Renamed to ...
15120 (vec_extract<mode><VEC_base_l>): ... this.
15121 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15122 (vec_initv2sfsf): ... this.
15123 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15124 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15125 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15126 element mode after mode in gen_vec_init* calls.
15127 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15128 (vec_init<mode><Vel>): ... this.
15129 (vec_extract<mode>): Renamed to ...
15130 (vec_extract<mode><Vel>): ... this.
15131 * config/aarch64/iterators.md (Vel): New mode attribute.
15132 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15133 Add element mode after mode in gen_vec_extract* calls.
15134 * config/s390/vector.md (non_vec_l): New mode attribute.
15135 (vec_extract<mode>): Renamed to ...
15136 (vec_extract<mode><non_vec_l>): ... this.
15137 (vec_init<mode>): Renamed to ...
15138 (vec_init<mode><non_vec_l>): ... this.
15139 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15140 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15142 * config/arm/iterators.md (V_elem_l): New mode attribute.
15143 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15144 (vec_extract<mode><V_elem_l>): ... this.
15145 (vec_extractv2di): Renamed to ...
15146 (vec_extractv2didi): ... this.
15147 (vec_init<mode>): Renamed to ...
15148 (vec_init<mode><V_elem_l>): ... this.
15149 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15150 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15151 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15152 Add element mode after gen_vec_extract* calls.
15153 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15154 (vec_init<mode><unitmode>): ... this.
15155 (vec_extract<mode>): Renamed to ...
15156 (vec_extract<mode><unitmode>): ... this.
15157 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15158 (vec_init<mode><unitmode>): ... this.
15159 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15160 (vec_initv2sfsf): ... this.
15161 (vec_extractv2sf): Renamed to ...
15162 (vec_extractv2sfsf): ... this.
15163 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15164 Add element mode after gen_vec_extract* calls.
15165 * config/mips/mips.md (unitmode): New mode iterator.
15166 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15167 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15168 * config/spu/spu.md (inner_l): New mode attribute.
15169 (vec_init<mode>): Renamed to ...
15170 (vec_init<mode><inner_l>): ... this.
15171 (vec_extract<mode>): Renamed to ...
15172 (vec_extract<mode><inner_l>): ... this.
15173 * config/sparc/sparc.md (veltmode): New mode iterator.
15174 (vec_init<VMALL:mode>): Renamed to ...
15175 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15176 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15177 (vec_initv2sisi): ... this.
15178 (vec_initv2sf): Renamed to ...
15179 (vec_initv2sfsf): ... this.
15180 (vec_extractv2sf): Renamed to ...
15181 (vec_extractv2sfsf): ... this.
15182 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15183 (vec_init<mode>): Renamed to ...
15184 (vec_init<mode><VEC_base_l>): ... this.
15185 (vec_extract<mode>): Renamed to ...
15186 (vec_extract<mode><VEC_base_l>): ... this.
15187 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15188 (vec_initv2sfsf): ... this.
15189 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15190 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15191 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15192 gen_vec_init* calls.
15194 2017-08-01 Richard Biener <rguenther@suse.de>
15196 PR tree-optimization/81297
15197 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15198 TREE_OVERFLOW from INTEGER_CSTs.
15200 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15202 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15204 2017-07-31 Carl Love <cel@us.ibm.com>
15206 * config/rs6000/rs6000-c: Add support for built-in functions
15207 vector signed char vec_xl_be (signed long long, signed char *);
15208 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15209 vector signed int vec_xl_be (signed long long, signed int *);
15210 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15211 vector signed long long vec_xl_be (signed long long, signed long long *);
15212 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15213 vector signed short vec_xl_be (signed long long, signed short *);
15214 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15215 vector double vec_xl_be (signed long long, double *);
15216 vector float vec_xl_be (signed long long, float *);
15217 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15218 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15219 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15221 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15222 (altivec_expand_builtin): Add switch statement to call
15223 altivec_expand_xl_be for each builtin.
15224 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15225 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15226 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15227 __builtin_vsx_le_be_v16qi.
15228 * doc/extend.texi: Update the built-in documentation file for the
15229 new built-in functions.
15231 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15234 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15236 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15238 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15240 * config.gcc: Add z14.
15241 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15242 CPU model numbers for z13s and z14.
15243 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15245 * config/s390/s390-opts.h (enum processor_type): Rename
15246 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15247 * config/s390/s390.c (processor_table): Add field for CPU name to
15248 be passed to Binutils.
15249 (s390_asm_output_machine_for_arch): Use the new field in
15250 processor_table for Binutils.
15251 (s390_expand_builtin): Replace arch12 with z14.
15252 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15253 (s390_get_sched_attrmask): Likewise.
15254 (s390_get_unit_mask): Likewise.
15255 * config/s390/s390.opt: Add z14 to processor_type enum.
15257 2017-07-31 Martin Jambor <mjambor@suse.cz>
15260 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15261 regardless of optimization level.
15263 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15264 Martin Liska <mliska@suse.cz>
15266 * predict.def: Remove old comment and adjust probability.
15267 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15268 PREDICT statements.
15270 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15273 * config/i386/i386.c (ix86_function_naked): New function.
15274 (ix86_can_use_return_insn_p): Return false for naked functions.
15275 (ix86_expand_prologue): Skip prologue for naked functions.
15276 (ix86_expand_epilogue): Skip epilogue for naked functions
15277 and emit trap instruction.
15278 (ix86_warn_func_return): New function.
15279 (ix86_attribute_table): Add "naked" attribute specification.
15280 (TARGET_WARN_FUNC_RETURN): Define.
15281 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15283 2017-07-31 Martin Liska <mliska@suse.cz>
15285 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15286 (dump_gimple_bb_header): Always dump BB info.
15287 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15289 2017-07-31 Martin Liska <mliska@suse.cz>
15292 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15293 also with current_function_decl non-null equality.
15295 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15298 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15299 change type to the element type, instead add eltype variable and
15300 use it where we are interested in the element type.
15302 PR tree-optimization/81603
15303 * ipa-polymorphic-call.c
15304 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15305 offset arithmetic in offset_int, bail out if the resulting bit offset
15306 doesn't fit into shwi.
15308 2017-07-31 Martin Liska <mliska@suse.cz>
15310 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15311 (gimplify_save_expr): Fix comment.
15313 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15316 * config/i386/i386.c (ix86_function_arg): Update arguments for
15318 (ix86_compute_frame_layout): Set the initial stack offset to
15319 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15320 INCOMING_FRAME_SP_OFFSET.
15321 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15322 stack before exception handler returns.
15323 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15324 the 'ERROR_CODE' for exception handler.
15326 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15328 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15329 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15330 (ASM_OUTPUT_REG_POP): Ditto.
15331 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15332 instead of asm_fprintf to output pure string.
15334 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15336 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15337 to imported_module_or_decl hook.
15338 (debug_nothing_tree_tree_tree_bool): Remove.
15339 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15340 * debug.c (do_nothing_debug_hooks): Use
15341 debug_nothing_tree_tree_tree_bool_bool instead of
15342 debug_nothing_tree_tree_tree_bool.
15343 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15344 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15345 * sdbout.c (sdb_debug_hooks): Likewise.
15346 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15347 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15349 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15350 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15351 attribute, don't add anything.
15353 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15355 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15356 (fold_build2_stat_loc): Likewise.
15357 (fold_build3_stat_loc): Likewise.
15358 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15359 (fold_build1_loc): Remove macro.
15360 (fold_build2_loc): Likewise.
15361 (fold_build3_loc): Likewise.
15363 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15365 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15366 (gimple_build_debug_bind_source_stat): Likewise.
15367 * gimple.h (gimple_build_debug_bind): Remove macro.
15368 (gimple_build_debug_bind_source): Likewise.
15370 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15372 * bitmap.c (bitmap_alloc): Adjust.
15373 (bitmap_gc_alloc): Likewise.
15374 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15376 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15378 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15379 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15380 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15381 (bitmap_gc_alloc_stat): Likewise.
15382 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15384 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15386 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15387 * rtl.h (shallow_copy_rtx): Remove macro.
15389 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15391 * emit-rtl.c (gen_raw_REG): Adjust.
15392 * gengenrtl.c (gendef): Likewise.
15393 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15394 * rtl.h (rtx_alloc): Remove macro.
15396 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15398 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15399 (build_tree_list_stat): Likewise.
15400 * tree.h (build_tree_list): Remove macro.
15401 (build_tree_list_vec): Likewise.
15403 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15405 * tree.c (make_vector_stat): Remove _stat from name.
15406 (build_vector_stat): Likewise.
15407 * tree.h (make_vector_stat): Remove macro.
15408 (build_vector_stat): Likewise.
15410 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15412 * tree.h (build_var_debug_value): Remove prototype.
15414 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15416 * tree.c (tree_cons_stat): Remove _stat from name.
15417 * tree.h (tree_cons): Remove macro.
15419 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15421 * tree.c (build_vl_exp_stat): Remove _stat from name.
15422 * tree.h (build_vl_exp): Remove macro.
15424 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15426 * tree.c (build_decl_stat): Remove _stat from name.
15427 * tree.h (build_decl): Remove macro.
15429 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15431 * gimple.c (gimple_build_with_ops_stat): Adjust.
15432 (gimple_alloc_stat): Remove _stat from name.
15433 * gimple.h (gimple_alloc): Remove macro.
15435 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15437 * tree.c (make_tree_vec_stat): Remove _stat from name.
15438 (grow_tree_vec_stat): Likewise.
15439 * tree.h (make_tree_vec_stat): Adjust prototype.
15440 (grow_tree_vec_stat): Likewise.
15441 (make_tree_vec): Remove macro.
15442 (grow_tree_vec): Likewise.
15444 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15446 * fold-const.c (fold_build1_stat_loc): Adjust.
15447 (fold_build2_stat_loc): Likewise.
15448 (fold_build3_stat_loc): Likewise.
15449 * tree.c (build0_stat): Remove _stat from name.
15450 (build1_stat): Likewise.
15451 (build2_stat): Likewise.
15452 (build3_stat): Likewise.
15453 (build4_stat): Likewise.
15454 (build5_stat): Likewise.
15455 * tree.h (build1_loc): Remove macro, and rename _stat function
15457 (build2_loc): Likewise.
15458 (build3_loc): Likewise.
15459 (build4_loc): Likewise.
15460 (build5_loc): Likewise.
15462 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15464 * tree.c (make_int_cst_stat): Remove _stat from name.
15465 * tree.h (make_int_cst_stat): Adjust prototype.
15466 (make_int_cst): Remove macro.
15468 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15470 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15471 * tree.h (make_tree_binfo_stat): Adjust prototype.
15472 (make_tree_binfo): Remove.
15474 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15476 * tree.c (copy_node_stat): Rename to copy_node.
15477 (build_distinct_type_copy): Adjust.
15478 * tree.h (copy_node_stat): Adjust prototype.
15479 (copy_node): Remove macro.
15481 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15483 * tree.c (make_node_stat): rename to make_node.
15484 (build_tree_list_stat): Adjust.
15485 (build0_stat): Likewise.
15486 (build2_stat): Likewise.
15487 (build3_stat): Likewise.
15488 (build4_stat): Likewise.
15489 (build5_stat): Likewise.
15490 (build_decl_stat): Likewise.
15491 * tree.h (make_node_stat): Adjust prototype.
15492 (make_node): remove macro.
15494 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15496 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15497 (PPC_FEATURE2_SCV): Likewise.
15498 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15500 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15502 * config/aarch64/aarch64.c
15503 (aarch64_internal_mov_immediate): Add new special pattern.
15504 * config/aarch64/aarch64.md (*movdi_aarch64):
15505 Add reg/32bit const mov case.
15507 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15508 Richard Sandiford <richard.sandiford@linaro.org>
15510 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15511 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15512 Add integer and movi cases.
15513 (movi-split-hf-df-sf split, fp16): New.
15514 (enabled): Added TARGET_FP_F16INST.
15515 * config/aarch64/iterators.md (GPF_HF): New.
15516 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15518 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15520 * config/aarch64/aarch64.c
15521 (aarch64_simd_container_mode): Add prototype.
15522 (aarch64_expand_mov_immediate): Add HI support.
15523 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15524 (aarch64_can_const_movi_rtx_p): New.
15525 (aarch64_preferred_reload_class):
15526 Remove restrictions of using FP registers for certain SIMD operations.
15527 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15528 (aarch64_valid_floating_const): Add integer move validation.
15529 (aarch64_simd_imm_scalar_p): Remove.
15530 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15531 (aarch64_legitimate_constant_p): Expand list of supported cases.
15532 * config/aarch64/aarch64-protos.h
15533 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15534 (aarch64_reinterpret_float_as_int): New.
15535 (aarch64_simd_imm_scalar_p): Remove.
15536 * config/aarch64/constraints.md (Uvi): New.
15537 (Dd): Split into Ds and new Dd.
15538 * config/aarch64/aarch64.md (*movsi_aarch64):
15540 (*movdi_aarch64): Add SIMD mov case.
15542 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15544 * tree-predcom.c: (struct chain): Handle store-store chain in which
15545 stores for elimination only store loop invariant values.
15546 (execute_pred_commoning_chain): Ditto.
15547 (prepare_initializers_chain_store_elim): Ditto.
15548 (prepare_finalizers): Ditto.
15549 (is_inv_store_elimination_chain): New function.
15550 (initialize_root_vars_store_elim_1): New function.
15552 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15554 * tree-predcom.c: Revise general description of the pass.
15555 (enum chain_type): New enum type for store elimination.
15556 (struct chain): New field supporting store elimination.
15557 (struct component): Ditto.
15558 (dump_chain): Dump store-stores chain.
15559 (release_chain): Release resources.
15560 (split_data_refs_to_components): Compute and create component
15561 contains only stores for elimination.
15562 (get_chain_last_ref_at): New function.
15563 (make_invariant_chain): Initialization.
15564 (make_rooted_chain): Specify chain type in parameter and record it.
15565 (add_looparound_copies): Skip for store-stores chain.
15566 (determine_roots_comp): Compute type of chain and pass it to
15568 (initialize_root_vars_store_elim_2): New function.
15569 (finalize_eliminated_stores): New function.
15570 (remove_stmt): Handle store for elimination.
15571 (execute_pred_commoning_chain): Execute predictive commoning on
15572 store-store chains.
15573 (determine_unroll_factor): Skip unroll for store-stores chain.
15574 (prepare_initializers_chain_store_elim): New function.
15575 (prepare_initializers_chain): Hanlde store-store chain.
15576 (prepare_finalizers_chain, prepare_finalizers): New function.
15577 (tree_predictive_commoning_loop): Return integer value indicating
15578 if loop is unrolled or lcssa form is corrupted.
15579 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
15581 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15583 * tree-predcom.c (initialize_root): Delete.
15584 (execute_pred_commoning_chain): Initialize root vars and replace
15585 reference of non-combined chain directly, rather than call above
15588 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15590 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
15591 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
15593 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15595 * tree-predcom.c (struct chain): New field init_seq.
15596 (release_chain): Release init_seq.
15597 (prepare_initializers_chain): Record intialization stmts in above
15599 (insert_init_seqs): New function.
15600 (tree_predictive_commoning_loop): Call insert_init_seqs.
15602 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15604 * tree-predcom.c (determine_roots_comp): Skip trivial components.
15606 2017-07-28 Richard Biener <rguenther@suse.de>
15608 * match.pd: Remove superfluous :c.
15609 * genmatch.c (simplify::id): Add member.
15610 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
15612 (current_id): New global.
15613 (dt_node::parent): Move from ...
15614 (dt_operand::parent): ... here. Add for_id member.
15615 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
15616 (decision_tree::find_node): Relax order requirement when
15617 merging DT_TRUE nodes to ones inbetween the current simplify
15618 and the one we try to merge with. Add diagnostic whenever
15619 we need to enforce pattern order by not merging.
15620 (decision_tree::insert): Set current_id.
15621 (decision_tree::print_node): Dump parent node and for_id.
15622 (parser::last_id): Add member.
15623 (parser::push_simplify): Assign unique id.
15624 (parser::parser): Initialize last_id.
15626 2017-07-28 Martin Liska <mliska@suse.cz>
15629 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
15630 gimple_build_debug_bind.
15632 2017-07-28 Richard Biener <rguenther@suse.de>
15634 PR tree-optimization/81502
15635 * match.pd: Add pattern combining BIT_INSERT_EXPR with
15637 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
15639 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
15640 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
15641 for BIT_FIELD_REF args.
15642 * fold-const.c (make_bit_field_ref): Likewise.
15643 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
15645 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15648 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
15649 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
15650 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
15651 Or it into SANITIZER_UNDEFINED.
15652 * ubsan.c: Include gimple-fold.h and varasm.h.
15653 (ubsan_expand_ptr_ifn): New function.
15654 (instrument_pointer_overflow): New function.
15655 (maybe_instrument_pointer_overflow): New function.
15656 (instrument_object_size): Formatting fix.
15657 (pass_ubsan::execute): Call instrument_pointer_overflow
15658 and maybe_instrument_pointer_overflow.
15659 * internal-fn.c (expand_UBSAN_PTR): New function.
15660 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
15661 * sanitizer.def (__ubsan_handle_pointer_overflow,
15662 __ubsan_handle_pointer_overflow_abort): New builtins.
15663 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
15664 * internal-fn.def (UBSAN_PTR): New internal function.
15665 * opts.c (sanitizer_opts): Add pointer-overflow.
15666 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
15667 * fold-const.c (build_range_check): Compute pointer range check in
15668 integral type if pointer arithmetics would be needed. Formatting
15671 2017-07-28 Martin Liska <mliska@suse.cz>
15674 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
15675 parameters that are of a variable-length.
15677 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
15679 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
15681 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
15682 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
15683 (CRT_CALL_STATIC_FUNCTION): Likewise.
15684 (ASM_DEFAULT_SPEC): New define.
15685 (ASM_SPEC32): Likewise.
15686 (ASM_SPEC64): Likewise.
15687 (ASM_SPEC_COMMON): Likewise.
15688 (ASM_SPEC): Likewise.
15689 (INVALID_64BIT): Likewise.
15690 (LINK_OS_DEFAULT_SPEC): Likewise.
15691 (LINK_OS_SPEC32): Likewise.
15692 (LINK_OS_SPEC64): Likewise.
15693 (POWERPC_LINUX): Likewise.
15694 (PTRDIFF_TYPE): Likewise.
15695 (RESTORE_FP_PREFIX): Likewise.
15696 (RESTORE_FP_SUFFIX): Likewise.
15697 (SAVE_FP_PREFIX): Likewise.
15698 (SAVE_FP_SUFFIX): Likewise.
15699 (SIZE_TYPE): Likewise.
15700 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
15701 (TARGET_64BIT): Likewise.
15702 (TARGET_64BIT): Likewise.
15703 (TARGET_AIX): Likewise.
15704 (WCHAR_TYPE_SIZE): Likewise.
15705 (WCHAR_TYPE): Undefine.
15706 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
15707 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
15708 (CPP_OS_RTEMS_SPEC): Delete.
15709 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
15710 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
15712 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
15714 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15716 PR tree-optimization/81578
15717 * tree-parloops.c (build_new_reduction): Bail out if
15718 reduction_code isn't one of the standard OpenMP reductions.
15719 Move the details printing after that decision.
15721 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15723 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
15724 related to reload_in_progress.
15725 (splat_input_operand): Likewise.
15726 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
15728 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
15730 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
15731 (TARGET_INSTANTIATE_DECLS): Likewise.
15732 (legitimate_indexed_address_p): Delete reload_in_progress code.
15733 (rs6000_debug_legitimate_address_p): Likewise.
15734 (rs6000_eliminate_indexed_memrefs): Likewise.
15735 (rs6000_emit_le_vsx_store): Likewise.
15736 (rs6000_emit_move_si_sf_subreg): Likewise.
15737 (rs6000_emit_move): Likewise.
15738 (register_to_reg_type): Likewise.
15739 (rs6000_pre_atomic_barrier): Likewise.
15740 (rs6000_machopic_legitimize_pic_address): Likewise.
15741 (rs6000_allocate_stack_temp): Likewise.
15742 (rs6000_address_for_fpconvert): Likewise.
15743 (rs6000_address_for_altivec): Likewise.
15744 (rs6000_secondary_memory_needed_rtx): Delete function.
15745 (rs6000_check_sdmode): Likewise.
15746 (rs6000_alloc_sdmode_stack_slot): Likewise.
15747 (rs6000_instantiate_decls): Likewise.
15748 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15749 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15750 Delete reload_in_progress.
15751 (*vec_reload_and_plus_<mptrsize>): Likewise.
15752 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15753 (vsx_div_v2di): Likewise.
15754 (vsx_udiv_v2di): Likewise.
15756 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15758 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15759 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15760 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15761 (rs6000_debug_reg_global): Delete print of LRA status.
15762 (rs6000_option_override_internal): Delete dead LRA related code.
15763 (rs6000_lra_p): Delete function.
15764 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15766 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
15768 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
15769 * config/riscv/rtems.h: New file.
15771 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15772 Sudakshina Das <sudi.das@arm.com>
15774 * config/aarch64/aarch64.md
15775 (define_split for and<mode>3nr_compare): Move
15776 non aarch64_logical_operand to a register.
15777 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
15778 register immediate operand to a register.
15779 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
15781 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15783 PR middle-end/81564
15784 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
15786 2017-07-27 Richard Biener <rguenther@suse.de>
15788 PR tree-optimization/81573
15789 PR tree-optimization/81494
15790 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
15791 multi defuse cycle case.
15793 2017-07-27 Richard Biener <rguenther@suse.de>
15795 PR tree-optimization/81571
15796 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
15799 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
15801 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
15802 earlier and only if MASK_FPU is set. Adjust formatting.
15804 2017-07-27 Martin Liska <mliska@suse.cz>
15806 * opt-functions.awk: Add validation of value of Init.
15807 * optc-gen.awk: Pass new argument.
15809 2017-07-27 Martin Liska <mliska@suse.cz>
15811 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
15812 Fix wrong condition.
15814 2017-07-27 Martin Liska <mliska@suse.cz>
15816 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
15817 BBs and edges seen by autoFDO.
15819 2017-07-27 Richard Biener <rguenther@suse.de>
15821 PR tree-optimization/81502
15822 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15823 with incompatible but same sized type.
15824 (execute_update_addresses_taken): Likewise.
15826 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15828 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15829 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15831 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15834 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15835 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15836 Change s_operand to memory_operand.
15838 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15840 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15841 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15842 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15843 Emit instructions rather than returning an expression. Handle TFmode
15844 and KFmode by casting to TImode.
15845 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15846 (rs6000_emit_le_vsx_store): Likewise.
15847 * config/rs6000/vsx.md (VSX_TI): New iterator.
15848 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15849 (*vsx_le_undo_permute_<mode>): Likewise.
15850 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15851 emit the split sequence.
15852 (*vsx_le_perm_store_<mode>): Likewise.
15854 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15856 PR tree-optimization/81555
15857 PR tree-optimization/81556
15858 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15859 if true, force CHANGED for the recursive invocation.
15860 (reassociate_bb): Remember original length of ops array, pass
15861 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15863 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15864 attributes for noipa attribute. For naked attribute use
15865 lookup_attribute first before lookup_attribute_spec.
15866 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15868 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15869 for functions with noipa attribute.
15870 (cgraph_externally_visible_p): Return true for functions with noipa
15872 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15873 for functions with noipa attribute.
15874 * doc/extend.texi: Document noipa function attribute.
15875 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15876 also for functions with noipa attribute.
15877 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15879 2017-07-26 Andrew Pinski <apinski@cavium.com>
15881 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15882 vec_unalign_load_cost and vec_unalign_store_cost.
15884 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15886 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15887 -mvsx-small-integer option.
15888 (ISA_3_0_MASKS_IEEE): Likewise.
15889 (OTHER_VSX_VECTOR_MASKS): Likewise.
15890 (POWERPC_MASKS): Likewise.
15891 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15892 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15893 code, only testing for DImode being allowed in non-VSX floating
15895 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15896 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15898 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15899 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15900 TARGET_P8_VECTOR test.
15901 (rs6000_secondary_reload_simple_move): Likewise.
15902 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15903 since TARGET_P9_VECTOR was already tested.
15904 (rs6000_opt_masks): Remove -mvsx-small-integer.
15905 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15906 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15908 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15909 test for TARGET_VEXTRACTUB was used, and that uses
15911 (p9 extract splitter): Likewise.
15912 (vsx_extract_<mode>_di_p9): Likewise.
15913 (vsx_extract_<mode>_store_p9): Likewise.
15914 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15915 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15916 the elimination of TARGET_VSX_SMALL_INTEGER.
15917 (vsx_extract_<mode>_p8): Likewise.
15918 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15919 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15920 (vsx_set_<mode>_p9): Likewise.
15921 (vsx_set_v4sf_p9): Likewise.
15922 (vsx_set_v4sf_p9_zero): Likewise.
15923 (vsx_insert_extract_v4sf_p9): Likewise.
15924 (vsx_insert_extract_v4sf_p9_2): Likewise.
15925 * config/rs6000/rs6000.md (sign extend splitter): Change
15926 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15927 (floatsi<mode>2_lfiwax_mem): Likewise.
15928 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15929 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15930 since a test for TARGET_P9_VECTOR was used.
15931 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15932 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15933 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15934 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15935 TARGET_P8_VECTOR test.
15936 (fix_trunc<mode>si2_stfiwx): Likewise.
15937 (fix_trunc<mode>si2_internal): Likewise.
15938 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15939 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15941 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15942 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15943 TARGET_P8_VECTOR test.
15944 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15945 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15946 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15948 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15949 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15950 since a test for TARGET_P9_VECTOR was used.
15951 (splitter for loading small constants): Likewise.
15953 2017-07-26 Andrew Pinski <apinski@cavium.com>
15955 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15958 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15961 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15962 (fp_valid_at): Likewise.
15964 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15966 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15967 (qdf24xx_addrcost_table): Likewise.
15968 (cortexa57_tunings): Update to use generic_branch_cost.
15969 (cortexa72_tunings): Likewise.
15970 (cortexa73_tunings): Likewise.
15971 (qdf24xx_tunings): Likewise.
15973 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15975 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15976 (thunderx2t99_branch_cost): Likewise.
15977 (cortexa35_tunings): Update to use generic_branch_cost.
15978 (cortexa53_tunings): Likewise.
15979 (cortexa57_tunings): Likewise.
15980 (cortexa72_tunings): Likewise.
15981 (cortexa73_tunings): Likewise.
15982 (thunderx2t99_tunings): Likewise.
15984 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15986 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15987 (sparc_option_override): Honour MASK_FSMULD.
15988 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15989 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15990 * config/sparc/sparc.opt (mfsmuld): New option.
15991 * doc/invoke.texi (mfsmuld): Document option.
15993 2017-07-26 Marek Polacek <polacek@redhat.com>
15995 PR middle-end/70992
15996 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15998 2017-07-26 Richard Biener <rguenther@suse.de>
16000 * gimple-match-head.c (do_valueize): Return OP if valueize
16002 (get_def): New helper to get at the def stmt of a SSA name
16003 if valueize allows.
16004 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16005 do_valueize to get at the def stmt.
16006 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16008 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16010 PR middle-end/46932
16011 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16013 2017-07-26 Martin Liska <mliska@suse.cz>
16016 * function.c (expand_function_start): Make expansion of
16017 nonlocal_goto_save_area after parm_birth_insn.
16019 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16021 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16022 from all CPU target flags enable members.
16024 2017-07-26 Richard Biener <rguenther@suse.de>
16026 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16027 (decision_tree::gen): Make 'type' const.
16028 (write_predicate): Likewise.
16030 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16032 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16033 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16034 (rs6000_option_override_internal): Likewise.
16035 (rs6000_expand_vector_set): Likewise.
16036 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16037 (TARGET_UPPER_REGS_SF): Likewise.
16038 (TARGET_UPPER_REGS_DI): Likewise.
16039 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16040 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16041 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16042 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16043 (Splitters for DI constants in Altivec registers): Likewise.
16044 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16045 (vsx_set_v4sf_p9): Likewise.
16046 (vsx_set_v4sf_p9_zero): Likewise.
16047 (vsx_insert_extract_v4sf_p9): Likewise.
16048 (vsx_insert_extract_v4sf_p9_2): Likewise.
16050 2017-07-25 Carl Love <cel@us.ibm.com>
16052 * doc/extend.texi: Update the built-in documentation file for the
16053 existing built-in functions
16054 vector signed char vec_cnttz (vector signed char);
16055 vector unsigned char vec_cnttz (vector unsigned char);
16056 vector signed short vec_cnttz (vector signed short);
16057 vector unsigned short vec_cnttz (vector unsigned short);
16058 vector signed int vec_cnttz (vector signed int);
16059 vector unsigned int vec_cnttz (vector unsigned int);
16060 vector signed long long vec_cnttz (vector signed long long);
16061 vector unsigned long long vec_cnttz (vector unsigned long long);
16063 2017-07-25 Andrew Pinski <apinski@cavium.com>
16065 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16066 accesses where the use is for the first operand of a BIT_INSERT.
16068 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16071 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16072 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16074 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16076 * config/i386/gstabs.h: Delete.
16077 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16079 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16081 * config/i386/i386.c (ix86_decompose_address): Do not check for
16082 register RTX when looking at index_reg or base_reg.
16083 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16085 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16087 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16088 to update EH info here.
16090 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16092 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16094 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16096 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16098 2017-07-25 Torsten Duwe <duwe@suse.de>
16100 * common.opt: Introduce -fpatchable-function-entry
16101 command line option, and its variables function_entry_patch_area_size
16102 and function_entry_patch_area_start.
16103 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16104 including a two-value parser.
16105 * target.def (print_patchable_function_entry): New target hook.
16106 * targhooks.h (default_print_patchable_function_entry): New function.
16107 * targhooks.c (default_print_patchable_function_entry): Likewise.
16108 * toplev.c (process_options): Switch off IPA-RA if
16109 patchable function entries are being generated.
16110 * varasm.c (assemble_start_function): Look at the
16111 patchable-function-entry command line switch and current
16112 function attributes and maybe generate NOP instructions by
16113 calling the print_patchable_function_entry hook.
16114 * doc/extend.texi: Document patchable_function_entry attribute.
16115 * doc/invoke.texi: Document -fpatchable_function_entry
16116 command line option.
16117 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16119 * doc/tm.texi: Re-generate.
16121 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16124 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16125 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16127 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16129 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16131 2017-07-25 Richard Biener <rguenther@suse.de>
16133 PR tree-optimization/81455
16134 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16135 not walk in cycles when looking for guards.
16137 2017-07-25 Richard Biener <rguenther@suse.de>
16139 PR tree-optimization/81529
16140 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16141 when optimizing backedge uses.
16143 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16145 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16147 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16148 to dl_section_ref. On AIX, append an expression to subtract
16149 the size of the section length to dl_section_ref.
16151 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16153 * configure.ac: If any of the config.* scripts fail, exit 1.
16154 * configure: Regenerate.
16156 2017-07-25 Richard Biener <rguenther@suse.de>
16158 PR middle-end/81546
16159 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16160 of immediate uses, be more verbose on errors.
16162 2017-07-25 Richard Biener <rguenther@suse.de>
16164 PR tree-optimization/81510
16165 * tree-vect-loop.c (vect_is_simple_reduction): When the
16166 reduction stmt is not inside the loop bail out.
16168 2017-07-25 Richard Biener <rguenther@suse.de>
16170 PR tree-optimization/81303
16171 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16172 profitability check against LOOP_VINFO_NITERSM1.
16174 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16176 * domwalk.c (cmp_bb_postorder): Simplify.
16177 (sort_bbs_postorder): New function. Use it...
16178 (dom_walker::walk): ...here to optimize common cases.
16180 2017-07-25 Martin Liska <mliska@suse.cz>
16183 * ipa-visibility.c (function_and_variable_visibility): Make the
16184 redirection just on target that supports aliasing.
16185 Fix GNU coding style.
16187 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16190 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16192 * config/arm/rtems.h: Likewise.
16193 * config/bfin/rtems.h: Likewise.
16194 * config/i386/rtemself.h: Likewise.
16195 * config/lm32/rtems.h: Likewise.
16196 * config/m32c/rtems.h: Likewise.
16197 * config/m68k/rtemself.h: Likewise.
16198 * config/microblaze/rtems.h: Likewise.
16199 * config/mips/rtems.h: Likewise.
16200 * config/moxie/rtems.h: Likewise.
16201 * config/nios2/rtems.h: Likewise.
16202 * config/powerpcspe/rtems.h: Likewise.
16203 * config/rs6000/rtems.h: Likewise.
16204 * config/rtems.h: Likewise.
16205 * config/sh/rtems.h: Likewise.
16206 * config/sh/rtemself.h: Likewise.
16207 * config/sparc/rtemself.h: Likewise.
16209 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16212 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16213 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16214 * tree-ssa-structalias.c (alias_get_name): Same.
16216 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16219 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16220 instructions if no du chain is found.
16222 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16224 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16226 2017-07-25 Richard Biener <rguenther@suse.de>
16228 PR middle-end/81505
16229 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16232 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16234 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16235 upper-regs options.
16236 (ISA_2_7_MASKS_SERVER): Likewise.
16237 (ISA_3_0_MASKS_IEEE): Likewise.
16238 (OTHER_P8_VECTOR_MASKS): Likewise.
16239 (OTHER_VSX_VECTOR_MASKS): Likewise.
16240 (POWERPC_MASKS): Likewise.
16241 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16242 duplicate list of options.
16243 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16244 explicit -mupper-regs options.
16245 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16246 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16247 alias for -mupper-regs-df.
16248 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16249 (rs6000_init_hard_regno_mode_ok): Likewise.
16250 (rs6000_option_override_internal): Likewise.
16251 (rs6000_opt_masks): Likewise.
16252 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16253 options in terms of whether -mvsx or -mpower8-vector was used.
16254 (TARGET_UPPER_REGS_DI): Likewise.
16255 (TARGET_UPPER_REGS_SF): Likewise.
16256 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16257 -mupper-regs-* options.
16259 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16261 * passes.c (emergency_dump_function): Print some empty lines and a
16262 header before the RTL dump.
16264 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16266 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16268 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16271 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16272 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16274 2017-07-24 Carl Love <cel@us.ibm.com>
16276 * config/rs6000/rs6000-c.c: Add support for built-in functions
16277 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16278 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16279 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16280 vec_extract_fp_from_shortl): Add defines for the two builtins.
16281 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16282 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16284 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16285 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16286 * doc/extend.texi: Update the built-in documentation file for the
16287 new built-in function.
16289 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16292 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16294 * doc/generic.texi: Likewise.
16295 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16296 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16298 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16300 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16301 (aarch64_mls_elt_merge<mode>): Likewise.
16303 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16305 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16306 having __cxa_atexit.
16308 2017-07-23 Michael Collison <michael.collison@arm.com>
16310 * config/arm/arm.c (arm_option_override): Deprecate
16311 use of -mstructure-size-boundary.
16312 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16313 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16315 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16318 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16319 Reduce cost estimate for direct moves.
16321 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16324 * config/i386/i386.c (ix86_option_override_internal): Disable
16325 BMI, BMI2 and TBM instructions for -m16.
16327 2017-07-21 Carl Love <cel@us.ibm.com>
16329 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16330 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16331 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16332 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16333 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16334 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16335 VMULOSW): New enum "unspec" values.
16336 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16337 altivec_vmulosw): New patterns.
16338 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16339 VMULOSW): Add definitions.
16341 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16343 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16344 (qdf24xx): Likewise.
16345 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16346 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16347 (AARCH64_FL_V8_1): Renumber.
16348 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16349 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16350 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16351 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16352 rdma to feature modifiers list.
16354 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16356 PR middle-end/56727
16357 * ipa-visibility (function_and_variable_visibility): Convert
16358 recursive PLT call to direct call if appropriate.
16360 2017-07-21 Andrew Pinski <apinski@cavium.com>
16362 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16363 operand 1 to see if the types precision matches.
16364 * fold-const.c (operand_equal_p): Likewise.
16366 2017-07-21 Richard Biener <rguenther@suse.de>
16368 PR tree-optimization/81303
16369 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16370 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16371 (vect_peeling_hash_get_lowest_cost): Adjust.
16372 (vect_enhance_data_refs_alignment): Likewise. Use
16373 vect_get_peeling_costs_all_drs to compute the penalty for no
16374 peeling to match up costs.
16376 2017-07-21 Richard Biener <rguenther@suse.de>
16378 PR tree-optimization/81500
16379 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16380 we didn't identify a reduction path.
16382 2017-07-21 Tom de Vries <tom@codesourcery.com>
16383 Cesar Philippidis <cesar@codesourcery.com>
16385 PR gcov-profile/81442
16386 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16389 2017-07-21 Tom de Vries <tom@codesourcery.com>
16392 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16394 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16395 nvptx_override_options_after_change.
16397 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16399 * dwarf2out.c (output_file_names): Avoid double testing for
16400 dwarf_version >= 5.
16402 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16404 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16406 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16408 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16410 (try_crossjump_to_edge): Do not punt on partitioned functions.
16412 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16414 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16415 Put all BBs reachable only via paths crossing cold region to cold
16417 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16419 2016-07-21 Richard Biener <rguenther@suse.de>
16421 PR tree-optimization/81303
16422 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16423 into account prologue and epilogue iterations when raising
16424 min_profitable_iters to sth at least covering one vector iteration.
16426 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16428 * config/arm/arm.c (arm_test_cpu_arch_dat):
16431 2017-07-20 Nathan Sidwell <nathan@acm.org>
16433 Remove TYPE_METHODS.
16434 * tree.h (TYPE_METHODS): Delete.
16435 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16436 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16437 (dbxout_type_methods): Scan TYPE_FIELDS.
16438 (dbxout_type): Don't check TYPE_METHODS here.
16439 * function.c (use_register_for_decl): Always ignore register for
16440 class types when not optimizing.
16441 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16442 * tree.c (free_lang_data_in_type): Stitch out member functions and
16443 templates from TYPE_FIELDS.
16444 (build_distinct_type_copy, verify_type_variant,
16445 verify_type): Member fns are on TYPE_FIELDS.
16446 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16447 * tree-pretty-print.c (dump_generic_node): Likewise.
16449 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16452 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16453 V2TImode and V4TImode.
16454 (ix86_expand_vector_extract): Likewise.
16455 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16456 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16457 (ssescalarmode): Handle V4TImode and V2TImode.
16458 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16459 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16460 (VEXTRACTI128_MODE): New mode iterator.
16461 (splitter for *vec_extractv?ti first element): New.
16462 (VEC_INIT_MODE): New mode iterator.
16463 (vec_init<mode>): Consolidate 3 expanders into one using
16464 VEC_INIT_MODE mode iterator.
16466 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16468 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16469 non_spilled_static_chain_regno_p.
16471 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16473 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16475 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16477 * bb-reorder.c (connect_traces): Allow copying of blocks within
16480 2017-07-20 Richard Biener <rguenther@suse.de>
16482 * gimple.h (gimple_phi_result): Add gphi * overload.
16483 (gimple_phi_result_ptr): Likewise.
16484 (gimple_phi_arg): Likewise. Adjust index assert to only
16485 allow actual argument accesses rather than all slots available
16487 (gimple_phi_arg_def): Add gphi * overload.
16488 * tree-phinodes.c (make_phi_node): Initialize only actual
16490 (resize_phi_node): Clear memory not covered by old node,
16491 do not initialize excess argument slots.
16492 (reserve_phi_args_for_new_edge): Initialize new argument slot
16495 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16497 PR tree-optimization/81388
16499 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16501 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16502 by removing computation of may_be_zero.
16504 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16505 Tom de Vries <tom@codesourcery.com>
16507 PR middle-end/81030
16508 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16509 when gimple level profile disagrees with what RTL expander did.
16511 2017-07-20 Richard Biener <rguenther@suse.de>
16513 PR tree-optimization/61171
16514 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16515 (vect_analyze_stmt): Add slp instance parameter.
16516 (vectorizable_reduction): Likewise.
16517 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16518 (vect_is_simple_reduction): Deal with chains not detected
16519 as SLP reduction chain, specifically not properly associated
16520 chains containing a mix of plus/minus.
16521 (get_reduction_op): Remove.
16522 (get_initial_defs_for_reduction): Simplify, pass in whether
16523 this is a reduction chain, pass in the SLP node for the PHIs.
16524 (vect_create_epilog_for_reduction): Get the SLP instance as
16526 (vectorizable_reduction): Get the SLP instance as arg.
16527 During analysis remember the SLP node with the PHIs in the
16528 instance. Simplify getting at the vectorized reduction PHIs.
16529 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16530 through SLP instance.
16531 (vect_slp_analyze_operations): Likewise.
16532 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16533 (vect_transform_stmt): Likewise.
16535 2017-07-20 Tom de Vries <tom@codesourcery.com>
16537 PR tree-optimization/81489
16538 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16539 read of phi arg location to before loop that modifies phi.
16541 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16543 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16546 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16548 PR middle-end/81331
16549 * except.c (execute): Fix ordering issue.
16551 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16553 PR rtl-optimization/81423
16554 * combine.c (make_compound_operation_int): Don't try to optimize
16555 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
16557 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16559 PR rtl-optimization/81423
16560 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
16561 with a constant that is -1 in the truncated to mode.
16563 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16565 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
16566 (determine_unlikely_bbs): ... here.
16567 * predict.h (propagate_unlikely_bbs_forward): Declare.
16568 * cfgexpand.c (pass_expand::execute): Use it.
16569 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
16571 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
16572 propagate_unlikely_bbs_forward.
16574 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16576 PR middle-end/81331
16577 * except.c (maybe_add_nop_after_section_switch): New function.
16580 2017-07-19 Tom de Vries <tom@codesourcery.com>
16582 * gimple.h (gimple_phi_set_arg): Make assert more strict.
16584 2017-07-19 Tom de Vries <tom@codesourcery.com>
16586 * gimple.h (gimple_phi_arg): Make assert more strict.
16588 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
16590 * config.gcc (powerpc*-*-*): Add mmintrin.h.
16591 * config/rs6000/mmintrin.h: New file.
16592 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
16594 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16596 PR tree-optimization/81346
16597 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
16599 2017-07-19 Tom de Vries <tom@codesourcery.com>
16601 * config/nvptx/nvptx.md (VECIM): Add V2DI.
16603 2017-07-19 Tom de Vries <tom@codesourcery.com>
16605 * config/nvptx/nvptx-modes.def: Add V2DImode.
16606 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
16607 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
16608 (nvptx_output_mov_insn): Handle lack of mov.b128.
16609 (nvptx_print_operand): Handle 'H' and 'L' codes.
16610 (nvptx_vector_mode_supported): Allow V2DImode.
16611 (nvptx_preferred_simd_mode): New function.
16612 (nvptx_data_alignment): New function.
16613 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
16614 nvptx_preferred_simd_mode.
16615 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
16617 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
16619 2017-07-19 Tom de Vries <tom@codesourcery.com>
16621 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
16622 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
16623 (nvptx_vector_mode_supported): New function. Allow V2SImode.
16624 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
16625 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
16626 (mov<VECIM>_insn): New define_insn.
16627 (define_expand "mov<VECIM>): New define_expand.
16629 2017-07-19 Tom de Vries <tom@codesourcery.com>
16631 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
16633 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16635 PR tree-optimization/81346
16636 * fold-const.h (fold_div_compare, range_check_type): Declare.
16637 * fold-const.c (range_check_type): New function.
16638 (build_range_check): Use range_check_type.
16639 (fold_div_compare): No longer static, rewritten into
16640 a match.pd helper function.
16641 (fold_comparison): Don't call fold_div_compare here.
16642 * match.pd (X / C1 op C2): New optimization using fold_div_compare
16643 as helper function.
16645 2017-07-19 Nathan Sidwell <nathan@acm.org>
16647 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
16648 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
16649 * tree.c (find_decls_types_r, verify_type): Use
16650 TYPE_{MIN,MAX}_VALUE_RAW.
16651 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
16652 (hash_tree): Likewise.
16653 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16655 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16658 2017-07-18 Tom de Vries <tom@codesourcery.com>
16660 PR middle-end/81464
16661 * omp-expand.c (expand_omp_for_static_chunk): Handle
16662 equal-argument loop exit phi.
16664 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
16667 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
16668 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
16669 operand 2 predicate.
16670 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
16671 operand 2 predicate.
16672 (ror,rol -> rorx splitters): Use const_int_operand as
16673 operand 2 predicate.
16675 2017-06-18 Richard Biener <rguenther@suse.de>
16677 PR tree-optimization/81410
16678 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
16679 the gap in the ! slp_perm SLP case after each group.
16681 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16683 PR middle-end/81463
16684 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
16687 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16689 PR middle-end/81462
16690 * predict.c (set_even_probabilities): Cleanup; do not affect
16691 probabilities that are already known.
16692 (combine_predictions_for_bb): Call even when count is set.
16694 2017-07-18 Nathan Sidwell <nathan@acm.org>
16696 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
16699 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16702 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
16703 optimization for loop niter analysis.
16705 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
16708 * config/avr/avr.c (avr_optimize_casesi): Don't use
16709 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
16711 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
16713 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
16714 body_cost_vec from _vect_peel_extended_info.
16715 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
16716 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
16719 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16721 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
16723 2017-07-18 Richard Biener <rguenther@suse.de>
16725 PR tree-optimization/80620
16726 PR tree-optimization/81403
16727 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
16728 info when re-using a VN table entry.
16730 2017-07-18 Richard Biener <rguenther@suse.de>
16732 PR tree-optimization/81418
16733 * tree-vect-loop.c (vectorizable_reduction): Properly compute
16734 vectype_in. Verify that with lane-reducing reduction operations
16735 we have a single def-use cycle.
16737 2017-07-17 Carl Love <cel@us.ibm.com>
16739 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16741 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16742 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16743 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16744 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16745 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16746 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16747 VMULOSW): New enum "unspec" values.
16748 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16749 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16750 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16751 altivec_vmulosw): New patterns.
16752 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16753 VMULOSW): Add definitions.
16754 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16756 * config/alpha/alpha.c: Include predict.h.
16758 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16760 * tree-vrp.c (compare_assert_loc): Fix comparison function
16761 to return predictable results.
16763 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16765 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16767 (subdi3): Likewise.
16768 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
16769 * doc/invoke.texi (mexpand-adddi): Update text.
16771 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16773 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
16774 that also clobbers the CC register. The old expand code is moved
16776 (*arc_clzsi2): ... here.
16777 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
16778 the CC register. The old expand code is moved to ...
16779 (arc_ctzsi2): ... here.
16781 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16783 * config/arc/arc.opt (mindexed-loads): Use initial value
16784 TARGET_INDEXED_LOADS_DEFAULT.
16785 (mauto-modify-reg): Use initial value
16786 TARGET_AUTO_MODIFY_REG_DEFAULT.
16787 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16788 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16789 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16790 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16792 2017-07-17 Martin Liska <mliska@suse.cz>
16795 * opts.c (finish_options): Do not allow -fgnu-tm
16796 w/ -fsanitize={kernel-,}address. Say sorry.
16798 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16801 * tree-loop-distribution.c (classify_partition): Only assert on
16802 numer of iterations.
16803 (merge_dep_scc_partitions): Delete prameter. Update function call.
16804 (distribute_loop): Remove code handling loop with unknown niters.
16805 (pass_loop_distribution::execute): Skip loop with unknown niters.
16807 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16810 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
16811 function sort_partitions_by_post_order.
16813 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16815 PR tree-optimization/81374
16816 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
16817 the max index of basic blocks, rather than number of basic blocks.
16819 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16821 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16823 (arc_legitimate_pic_operand_p): Likewise.
16824 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16826 (arc_needs_pcl_p): Likewise.
16827 (arc_legitimate_pc_offset_p): Likewise.
16828 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16829 function is also used in constrains.md.
16830 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16831 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16832 PLUS. Only return true/false in known cases, otherwise assert.
16833 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16834 is already called in arc_legitimate_constant_p.
16835 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16837 (LEGITIMATE_PIC_OPERAND_P): Use
16838 arc_raw_symbolic_reference_mentioned_p function.
16839 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16844 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16845 Andrew Burgess <andrew.burgess@embecosm.com>
16847 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16848 (arc_return_address_register): New function.
16849 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16850 (arc_handle_fndecl_attribute): Add naked attribute.
16851 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16852 (TARGET_WARN_FUNC_RETURN): Likewise.
16853 (arc_allocate_stack_slots_for_args): New function.
16854 (arc_warn_func_return): Likewise.
16855 (machine_function): Change type fn_type.
16856 (arc_compute_function_type): Consider new naked function type,
16857 change function return type.
16858 (arc_must_save_register): Adapt to handle new
16859 arc_compute_function_type's return type.
16860 (arc_expand_prologue): Likewise.
16861 (arc_expand_epilogue): Likewise.
16862 (arc_return_address_regs): Delete.
16863 (arc_return_address_register): New function.
16864 (arc_epilogue_uses): Use above function.
16865 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16866 (arc_function_type): Change encoding, add naked type.
16867 (ARC_INTERRUPT_P): Change to handle the new encoding.
16868 (ARC_FAST_INTERRUPT_P): Likewise.
16869 (ARC_NORMAL_P): Define.
16870 (ARC_NAKED_P): Likewise.
16871 (arc_compute_function_type): Delete prototype.
16872 * config/arc/arc.md (in_ret_delay_slot): Use
16873 arc_return_address_register function.
16874 (simple_return): Likewise.
16875 (p_return_i): Likewise.
16877 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16879 PR tree-optimization/81428
16880 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16881 can't be built for those types.
16883 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16885 Remove stuff dead since r239246.
16887 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16888 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16889 (avr_inform_devices): Remove dead stuff.
16891 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16893 * config/arm/arm_neon.h: Fix softp typo.
16895 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16897 PR tree-optimization/81365
16898 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16899 aggregate moves onto bb predecessor edges, make sure there are no
16900 loads that could alias the lhs in between the start of bb and the
16903 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16906 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16907 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16908 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16910 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16912 PR tree-optimization/81396
16913 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16914 (init_symbolic_number): Initialize it to 1.
16915 (perform_symbolic_merge): Add n_ops from both operands into the new
16917 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16918 without base_addr as useless if they need more than one operation.
16919 (bswap_replace): Handle !bswap case for NULL base_addr.
16921 2017-07-17 Tom de Vries <tom@codesourcery.com>
16924 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16927 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16929 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16930 conditional builtin define __FIX_LEON3FT_B2BST.
16932 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16934 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16935 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16938 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16940 PR rtl-optimization/81424
16941 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16942 to remove potential trapping from operands if -fnon-call-exceptions.
16944 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16946 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16947 profile_proability for scalling.
16948 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16950 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16952 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16954 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16956 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16957 fixpoint arithmetics.
16959 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16961 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16962 fixpoint arithmetics.
16964 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16966 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16967 fixpoint arithmetics.
16969 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16971 * profile-count.h (profile_probability::from_reg_br_prob_note,
16972 profile_probability::to_reg_br_prob_note): New functions.
16973 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16974 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16975 * predict.c (probability_reliable_p): Update.
16976 (edge_probability_reliable_p): Update.
16977 (br_prob_note_reliable_p): Update.
16978 (invert_br_probabilities): Update.
16979 (add_reg_br_prob_note): New function.
16980 (combine_predictions_for_insn): Update.
16981 * asan.c (asan_clear_shadow): Update.
16982 * cfgbuild.c (compute_outgoing_frequencies): Update.
16983 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16984 (update_br_prob_note): Update.
16985 (rtl_verify_edges): Update.
16986 (purge_dead_edges): Update.
16987 (fixup_reorder_chain): Update.
16988 * emit-rtl.c (try_split): Update.
16989 * ifcvt.c (cond_exec_process_insns): Update.
16990 (cond_exec_process_if_block): Update.
16991 (dead_or_predicable): Update.
16992 * internal-fn.c (expand_addsub_overflow): Update.
16993 (expand_neg_overflow): Update.
16994 (expand_mul_overflow): Update.
16995 * loop-doloop.c (doloop_modify): Update.
16996 * loop-unroll.c (compare_and_jump_seq): Update.
16997 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16998 * predict.h: Update.
16999 * reorg.c (mostly_true_jump): Update.
17001 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17002 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17003 * config/arc/arc.c: (emit_unlikely_jump): Update.
17004 * config/arm/arm.c: (emit_unlikely_jump): Update.
17005 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17006 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17007 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17008 (ix86_print_operand): Update.
17009 (ix86_split_fp_branch): Update.
17010 (predict_jump): Update.
17011 * config/ia64/ia64.c (ia64_print_operand): Update.
17012 * config/mmix/mmix.c (mmix_print_operand): Update.
17013 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17014 (rs6000_expand_split_stack_prologue): Update.
17015 * config/rs6000/rs6000.c: Update.
17016 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17017 (s390_expand_vec_movstr): Update.
17018 (s390_expand_cs_tdsi): Update.
17019 (s390_expand_split_stack_prologue): Update.
17020 * config/sh/sh.c (sh_print_operand): Update.
17021 (expand_cbranchsi4): Update.
17022 (expand_cbranchdi4): Update.
17023 * config/sparc/sparc.c (output_v9branch): Update.
17024 * config/spu/spu.c (get_branch_target): Update.
17025 (ea_load_store_inline): Update.
17026 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17027 * config/tilepro/tilepro.c: Update.
17029 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17031 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17032 (gimplify_save_expr): Likewise.
17034 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17036 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17038 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17040 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17042 * timevar.def (TV_IPA_FNSUMMARY): Define.
17044 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17046 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17047 to back store errata sensitive sequence from being generated.
17048 (sqrtdf2_fix): Likewise.
17050 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17052 * tree-ssa-threadupdate.c (compute_path_counts,
17053 update_joiner_offpath_counts): Use profile_probability.
17055 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17058 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17060 * config/arm/arm-c.c (arm_cpu_builtins): Define
17061 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17063 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17065 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17066 array entries to represent __ieee128 versions of the
17067 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17068 scalar_extract_sig, and scalar_insert_exp built-in functions.
17069 (altivec_resolve_overloaded_builtin): Add special case handling
17070 for the __builtin_scalar_insert_exp function, as represented by
17071 the P9V_BUILTIN_VEC_VSIEDP constant.
17072 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17073 exponent support for __ieee128 argument.
17074 (VSESQP): Add scalar extract signature support for __ieee128
17076 (VSTDCNQP): Add scalar test negative support for __ieee128
17078 (VSIEQP): Add scalar insert exponent support for __int128 argument
17079 with __ieee128 result.
17080 (VSIEQPF): Add scalar insert exponent support for __ieee128
17081 argument with __ieee128 result.
17082 (VSTDCQP): Add scalar test data class support for __ieee128
17084 (VSTDCNQP): Add overload support for scalar test negative with
17085 __ieee128 argument.
17086 (VSTDCQP): Add overload support for scalar test data class
17087 __ieee128 argument.
17088 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17089 UNSPEC_VSX_SXSIGDP.
17090 (UNSPEC_VSX_SIEXPQP): New constant.
17091 (xsxexpqp): New insn for VSX scalar extract exponent quad
17093 (xsxsigqp): New insn for VSX scalar extract significand quad
17095 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17096 precision with floating point argument.
17097 (xststdcqp): New expand for VSX scalar test data class quad
17099 (xststdcnegqp): New expand for VSX scalar test negative quad
17101 (xststdcqp): New insn to match expansions for VSX scalar test data
17102 class quad precision and VSX scalar test negative quad precision.
17103 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17104 special case operand checking to enforce that second operand of
17105 VSX scalar test data class with quad precision argument is a 7-bit
17107 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17108 prototypes and descriptions of __ieee128 versions of
17109 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17110 scalar_test_data_class, and scalar_test_neg built-in functions.
17112 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17114 PR tree-optimization/81162
17115 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17116 replace a negate with an add.
17118 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17120 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17122 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17124 * config/arm/arm-c.c (arm_cpu_builtins): Define
17125 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17127 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17129 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17130 (armv8-r): Set ARM Cortex-R52 as default CPU.
17131 * config/arm/arm-tables.opt: Regenerate.
17132 * config/arm/arm-tune.md: Regenerate.
17133 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17135 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17136 extension for -mcpu=cortex-r52.
17138 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17140 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17141 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17142 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17143 (fp-armv8): Define it as FP_ARMv8 only.
17144 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17145 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17147 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17148 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17149 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17150 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17151 than TARGET_FPU_ARMV8.
17152 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17153 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17154 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17156 * config/arm/neon.md (neon_vrint): Likewise.
17157 (neon_vcvt): Likewise.
17158 (neon_<fmaxmin_op><mode>): Likewise.
17159 (<fmaxmin><mode>3): Likewise.
17160 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17161 * config/arm/predicates.md (arm_cond_move_operator): Check against
17162 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17164 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17166 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17167 to top of function.
17169 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17171 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17172 loop in comment with memset.
17174 2017-07-14 Martin Liska <mliska@suse.cz>
17176 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17177 * dwarf2out.c (is_java): Remove the function.
17178 (output_pubname): Remove usage of the function.
17179 (lower_bound_default): Remove usage of DW_LANG_Java.
17180 (gen_compile_unit_die): Likewise.
17181 * gcc.c: Remove compiler defaults for .java and .zip files.
17182 * gimple-expr.c (remove_suffix): Change as there's no longer
17183 extension than 4-letter one.
17184 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17185 (gimplify_save_expr): Likewise.
17186 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17187 as it's possible even for other languages than Java.
17188 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17189 * lto-opts.c (lto_write_options): Remove reference to Java.
17190 * opts.c (strip_off_ending): Update file extension handling.
17191 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17192 * tree-eh.c (lower_resx): Likewise.
17193 * tree.c (free_lang_data_in_type): Remove dead code.
17194 (find_decls_types_r): Likewise.
17195 (build_common_builtin_nodes): Remove Java from a comment.
17196 (verify_type): Remove dead code.
17197 * varasm.c (assemble_external): Remove Java from a comment.
17199 2017-07-14 Martin Liska <mliska@suse.cz>
17201 * opts.c (finish_options): Add quotes.
17202 (common_handle_option): Likewise.
17204 2017-07-14 Martin Liska <mliska@suse.cz>
17206 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17207 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17208 Remove N_SO_PASCAL.
17209 * dwarf2out.c (lower_bound_default): Do not handle
17211 (gen_compile_unit_die): Likewise.
17212 * gcc.c: Remove default extension binding for GNU Pascal.
17213 * stmt.c: Remove Pascal language from a comment.
17214 * xcoffout.c: Likewise.
17216 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17219 * diagnostic-show-locus.c (fixit_cmp): New function.
17220 (layout::layout): Sort m_fixit_hints.
17221 (column_range::column_range): Assert that the values are valid.
17222 (struct char_span): New struct.
17223 (correction::overwrite): New method.
17224 (struct source_line): New struct.
17225 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17226 calls in terms of classes source_line and char_span, and
17227 correction::overwrite.
17228 (selftest::test_overlapped_fixit_printing_2): New function.
17229 (selftest::diagnostic_show_locus_c_tests): Call it.
17231 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17233 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17234 early if there is no lhs.
17236 2017-07-13 Martin Liska <mliska@suse.cz>
17238 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17239 (gen_reference_type_die): Likewise.
17240 * stor-layout.c: Remove Pascal-related comment.
17242 2017-07-13 Martin Liska <mliska@suse.cz>
17244 * opts.c (finish_options): Add quotes to error messages.
17245 (parse_sanitizer_options): Likewise.
17247 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17249 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17251 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17253 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17255 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17257 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17259 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17261 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17264 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17265 provides the hardware capability bits, define the macro
17266 __BUILTIN_CPU_SUPPORTS__.
17267 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17268 if GLIBC does not provide the hardware capability bits. Add a
17269 gcc_unreachable call if the built-in cpu function is neither
17270 __builtin_cpu_is nor __builtin_cpu_supports.
17271 (rs6000_get_function_versions_dispatcher): Change the warning
17272 that an old GLIBC is used which does not export the capability
17273 bits to be an error.
17274 * doc/extend.texi (target_clones attribute): Document the
17275 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17276 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17277 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17278 the macros defined by GCC if the newer GLIBC is available.
17280 2017-07-12 Jeff Law <law@redhat.com>
17282 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17283 remaining includes slightly.
17284 * config/riscv/riscv-builtins.c: Include profile-count.h.
17286 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17289 * config/avr/avr.c (avr_set_current_function): In diagnostic
17290 messages: Quote keywords and (parts of) identifiers.
17291 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17294 2017-07-12 Carl Love <cel@us.ibm.com>
17296 * config/rs6000/rs6000-c.c: Add support for built-in functions
17297 vector bool char vec_revb (vector bool char);
17298 vector bool short vec_revb (vector short char);
17299 vector bool int vec_revb (vector bool int);
17300 vector bool long long vec_revb (vector bool long long);
17301 * doc/extend.texi: Update the built-in documentation file for the
17302 new built-in functions.
17304 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17306 * config/s390/s390.md: Remove movcc splitter.
17308 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17310 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17311 load/store on condition.
17313 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17316 * config/avr/avr.c (avr_encode_section_info)
17317 [progmem && !TREE_READONLY]: Error if progmem object needs
17320 2017-07-11 Michael Collison <michael.collison@arm.com>
17322 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17325 2017-07-11 Carl Love <cel@us.ibm.com>
17327 * config/rs6000/rs6000-c.c: Add support for builtins
17328 vector unsigned int vec_parity_lsbb (vector signed int);
17329 vector unsigned int vec_parity_lsbb (vector unsigned int);
17330 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17331 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17332 vector unsigned long long vec_parity_lsbb (vector signed long long);
17333 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17334 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17335 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17336 * doc/extend.texi: Update the built-in documentation file for the
17337 new built-in functions.
17339 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17341 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17342 (layout::m_primary_loc): New field.
17343 (layout::layout): Initialize new field. Move location filtering
17344 logic from here to...
17345 (layout::maybe_add_location_range): ...this new method. Add
17346 support for filtering to just the lines already specified by other
17348 (layout::will_show_line_p): New method.
17349 (gcc_rich_location::add_location_if_nearby): New method.
17350 (selftest::test_add_location_if_nearby): New test function.
17351 (selftest::diagnostic_show_locus_c_tests): Call it.
17352 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17355 2017-07-11 Tom de Vries <tom@codesourcery.com>
17357 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17358 (bb_first_real_insn): New function.
17359 (nvptx_single): Add extra initialization of broadcasted condition
17362 2017-07-11 Nathan Sidwell <nathan@acm.org>
17364 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17366 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17368 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17369 Binutils doc as TEXI will mess them up.
17370 * doc/invoke.texi (AVR Options): Same here.
17372 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17374 * config/sparc/sparc.opt (mfix-ut700): New option.
17375 (mfix-gr712rc): Likewise.
17376 (sparc_fix_b2bst): New variable.
17377 * doc/invoke.texi (SPARC options): Document them.
17378 (ARM options): Fix warnings.
17379 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17380 instructions to prevent sequences that can trigger the store-store
17381 errata for certain LEON3FT processors.
17382 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17383 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17384 * config/sparc/sparc.md (fix_b2bst): New attribute.
17385 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17387 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17390 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17392 (*rsqrtsf2_sse): Ditto.
17394 (div<mode>3): Macroize insn from divdf3 and divsf3
17395 using MODEF mode iterator.
17397 2017-07-10 Martin Sebor <msebor@redhat.com>
17399 PR tree-optimization/80397
17400 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17401 instead of testing for equality to INTEGER_TYPE.
17403 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17405 * config.gcc: Remove uclibc from arc target spec.
17407 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17409 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17411 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17414 * lto-wrapper.c (remove_option): New function.
17415 (merge_and_complain): Merge PIC/PIE options more realistically.
17417 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17419 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17423 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17424 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17425 * config.in: Regenerate.
17426 * configure: Regenerate.
17427 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17428 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17429 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17430 (TARGET_GASISR_PROLOGUES): ...target mask.
17431 * common/config/avr/avr-common.c
17432 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17433 Set -mgas-isr-prologues.
17434 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17435 INSERT_PASS_BEFORE for it.
17436 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17437 * config/avr/avr.c (avr_option_override)
17438 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17439 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17440 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17441 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17442 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17444 (make_avr_pass_pre_proep): New function.
17445 (emit_push_sfr) <treg>: Add argument to function and use it
17446 instead of TMP_REG.
17447 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17448 and set machine->gasisr.yes.
17449 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17450 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17451 __gcc_isr.n_pushed to .L__stack_usage.
17452 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17453 (avr_asm_final_postscan_insn): ...this new static function.
17454 * config/avr/avr.h (machine_function)
17455 <is_no_gccisr, use_L__stack_usage>: New fields.
17456 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17457 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17458 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17459 (gasisr, *gasisr): New expander and insn.
17460 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17461 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17462 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17464 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17466 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17469 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17471 Move jump-tables out of .text again.
17474 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17475 (ASM_OUTPUT_ADDR_VEC): New function.
17476 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17477 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17478 INSN_ADDRESSes as asm comment.
17479 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17480 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17481 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17482 * config/avr/avr.md (*tablejump): Adjust comment.
17483 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17484 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17486 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17487 (avr_output_addr_vec): New proto.
17488 (avr_log_t) <insn_addresses>: New field.
17490 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17493 * config/i386/i386.c (ix86_function_arg_advance): Set
17494 outgoing_args_on_stack to true if there are outgoing arguments
17496 (ix86_function_arg): Likewise.
17497 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17498 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17499 * config/i386/i386.h (machine_function): Add
17500 outgoing_args_on_stack.
17502 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17504 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17505 supporting pthreds.
17506 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17508 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17510 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17511 (REAL_H): Remove $(MACHMODE_H).
17512 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17514 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17515 $(MACHMODE_H) and double-int.h.
17516 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17518 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17521 2017-07-07 Andrew Pinski <apinski@cavium.com>
17523 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17524 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17526 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17528 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17529 Add warning if GCC was not configured to link against a GLIBC that
17530 exports the hardware capability bits.
17531 (make_resolver_func): Make resolver function private and not a
17532 COMDAT function. Create the name with clone_function_name instead
17533 of make_unique_name.
17536 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17537 correct operand in doing the split.
17539 2017-07-07 Carl Love <cel@us.ibm.com>
17541 * config/rs6000/rs6000-c: Add support for built-in function
17542 vector unsigned short vec_pack_to_short_fp32 (vector float,
17544 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17545 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17546 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17547 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17548 (convert_4f32_8i16): Add define_expand.
17549 * doc/extend.texi: Update the built-in documentation file for the
17550 new built-in function.
17552 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17554 * config/sparc/m8.md: New file.
17555 * config/sparc/sparc.md: Include m8.md.
17557 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17559 * config/sparc/sparc.opt: New option -mvis4b.
17560 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
17561 (sparc_option_override): Handle VIS4B.
17562 (enum sparc_builtins): Define
17563 SPARC_BUILTIN_DICTUNPACK{8,16,32},
17564 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
17565 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
17566 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
17567 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
17568 (check_constant_argument): New function.
17569 (sparc_vis_init_builtins): Define builtins
17570 __builtin_vis_dictunpack{8,16,32},
17571 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
17572 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
17573 __builtin_vis_fpcmpde{8,16,32}shl and
17574 __builtin_vis_fpcmpur{8,16,32}shl.
17575 (sparc_expand_builtin): Check that the constant operands to
17576 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
17577 constant and in range.
17578 * config/sparc/sparc-c.c (sparc_target_macros): Handle
17580 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
17581 (SPARC_IMM5_P): Likewise.
17582 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
17583 (enabled): Handle vis4b.
17584 (UNSPEC_DICTUNPACK): New unspec.
17585 (UNSPEC_FPCMPSHL): Likewise.
17586 (UNSPEC_FPUCMPSHL): Likewise.
17587 (UNSPEC_FPCMPDESHL): Likewise.
17588 (UNSPEC_FPCMPURSHL): Likewise.
17589 (cpu_feature): New CPU feature `vis4b'.
17590 (dictunpack{8,16,32}): New insns.
17591 (FPCSMODE): New mode iterator.
17592 (fpcscond): New code iterator.
17593 (fpcsucond): Likewise.
17594 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
17595 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
17596 (fpcmpde{8,16,32}{si,di}shl): Likewise.
17597 (fpcmpur{8,16,32}{si,di}shl): Likewise.
17598 * config/sparc/constraints.md: Define constraints `q' for unsigned
17599 2-bit integer constants and `t' for unsigned 5-bit integer
17601 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
17603 (imm5_operand_dictunpack16): Likewise.
17604 (imm5_operand_dictunpack32): Likewise.
17605 (imm2_operand): Likewise.
17606 * doc/invoke.texi (SPARC Options): Document -mvis4b.
17607 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
17608 ditunpack* and fpcmp*shl builtins.
17610 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17612 * config.gcc: Handle m8 in --with-{cpu,tune} options.
17613 * config.in: Add HAVE_AS_SPARC6 define.
17614 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
17616 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
17618 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
17619 (CPP_CPU_SPEC): Handle m8.
17620 (ASM_CPU_SPEC): Likewise.
17621 * config/sparc/sparc-opts.h (enum processor_type): Add
17623 * config/sparc/sparc.c (m8_costs): New struct.
17624 (sparc_option_override): Handle TARGET_CPU_m8.
17625 (sparc32_initialize_trampoline): Likewise.
17626 (sparc64_initialize_trampoline): Likewise.
17627 (sparc_issue_rate): Likewise.
17628 (sparc_register_move_cost): Likewise.
17629 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
17630 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
17631 (ASM_CPU64_DEFAULT_SPEC): Likewise.
17632 (CPP_CPU_SPEC): Handle M8.
17633 (ASM_CPU_SPEC): Likewise.
17634 (AS_M8_FLAG): Define.
17635 * config/sparc/sparc.md: Add m8 to the cpu attribute.
17636 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
17637 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
17639 * configure: Regenerate.
17640 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
17643 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17645 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
17647 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
17648 ("*movdi_insn_sp32"): Do not set v3pipe.
17649 ("*movsi_insn"): Likewise.
17650 ("*movdi_insn_sp64"): Likewise.
17651 ("*movsf_insn"): Likewise.
17652 ("*movdf_insn_sp32"): Likewise.
17653 ("*movdf_insn_sp64"): Likewise.
17654 ("*zero_extendsidi2_insn_sp64"): Likewise.
17655 ("*sign_extendsidi2_insn"): Likewise.
17656 ("*mov<VM32:mode>_insn"): Likewise.
17657 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17658 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17659 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17660 ("<vlop:code><VL:mode>3"): Likewise.
17661 ("*not_<vlop:code><VL:mode>3"): Likewise.
17662 ("*nand<VL:mode>_vis"): Likewise.
17663 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
17664 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
17665 ("one_cmpl<VL:mode>2"): Likewise.
17666 ("faligndata<VM64:mode>_vis"): Likewise.
17667 ("alignaddrsi_vis"): Likewise.
17668 ("alignaddrdi_vis"): Likweise.
17669 ("alignaddrlsi_vis"): Likewise.
17670 ("alignaddrldi_vis"): Likewise.
17671 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17672 ("bmaskdi_vis"): Likewise.
17673 ("bmasksi_vis"): Likewise.
17674 ("bshuffle<VM64:mode>_vis"): Likewise.
17675 ("cmask8<P:mode>_vis"): Likewise.
17676 ("cmask16<P:mode>_vis"): Likewise.
17677 ("cmask32<P:mode>_vis"): Likewise.
17678 ("pdistn<P:mode>_vis"): Likewise.
17679 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17681 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17683 * config/sparc/sparc.md ("subtype"): New insn attribute.
17684 ("*wrgsr_sp64"): Set insn subtype.
17685 ("*rdgsr_sp64"): Likewise.
17686 ("alignaddrsi_vis"): Likewise.
17687 ("alignaddrdi_vis"): Likewise.
17688 ("alignaddrlsi_vis"): Likewise.
17689 ("alignaddrldi_vis"): Likewise.
17690 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17691 ("fexpand_vis"): Likewise.
17692 ("fpmerge_vis"): Likewise.
17693 ("faligndata<VM64:mode>_vis"): Likewise.
17694 ("bshuffle<VM64:mode>_vis"): Likewise.
17695 ("cmask8<P:mode>_vis"): Likewise.
17696 ("cmask16<P:mode>_vis"): Likewise.
17697 ("cmask32<P:mode>_vis"): Likewise.
17698 ("fchksm16_vis"): Likewise.
17699 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
17700 ("fmean16_vis"): Likewise.
17701 ("fp<plusminus_insn>64_vis"): Likewise.
17702 ("<plusminus_insn>v8qi3"): Likewise.
17703 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17704 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
17705 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
17706 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
17707 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
17708 ("*movqi_insn"): Likewise.
17709 ("*movhi_insn"): Likewise.
17710 ("*movsi_insn"): Likewise.
17711 ("movsi_pic_gotdata_op"): Likewise.
17712 ("*movdi_insn_sp32"): Likewise.
17713 ("*movdi_insn_sp64"): Likewise.
17714 ("movdi_pic_gotdata_op"): Likewise.
17715 ("*movsf_insn"): Likewise.
17716 ("*movdf_insn_sp32"): Likewise.
17717 ("*movdf_insn_sp64"): Likewise.
17718 ("*zero_extendhisi2_insn"): Likewise.
17719 ("*zero_extendqihi2_insn"): Likewise.
17720 ("*zero_extendqisi2_insn"): Likewise.
17721 ("*zero_extendqidi2_insn"): Likewise.
17722 ("*zero_extendhidi2_insn"): Likewise.
17723 ("*zero_extendsidi2_insn_sp64"): Likewise.
17724 ("ldfsr"): Likewise.
17725 ("prefetch_64"): Likewise.
17726 ("prefetch_32"): Likewise.
17727 ("tie_ld32"): Likewise.
17728 ("tie_ld64"): Likewise.
17729 ("*tldo_ldub_sp32"): Likewise.
17730 ("*tldo_ldub1_sp32"): Likewise.
17731 ("*tldo_ldub2_sp32"): Likewise.
17732 ("*tldo_ldub_sp64"): Likewise.
17733 ("*tldo_ldub1_sp64"): Likewise.
17734 ("*tldo_ldub2_sp64"): Likewise.
17735 ("*tldo_ldub3_sp64"): Likewise.
17736 ("*tldo_lduh_sp32"): Likewise.
17737 ("*tldo_lduh1_sp32"): Likewise.
17738 ("*tldo_lduh_sp64"): Likewise.
17739 ("*tldo_lduh1_sp64"): Likewise.
17740 ("*tldo_lduh2_sp64"): Likewise.
17741 ("*tldo_lduw_sp32"): Likewise.
17742 ("*tldo_lduw_sp64"): Likewise.
17743 ("*tldo_lduw1_sp64"): Likewise.
17744 ("*tldo_ldx_sp64"): Likewise.
17745 ("*mov<VM32:mode>_insn"): Likewise.
17746 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17747 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17749 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17751 * config/sparc/sparc.md ("type"): New insn type viscmp.
17752 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17754 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17755 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17756 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17757 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17759 ("n7_vis_logical_11cycle"): Likewise.
17760 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17761 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17762 * config/sparc/niagara.md ("niag_vis"): Likewise.
17763 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17764 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17766 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17768 * config/sparc/sparc.md: New instruction type `bmask'.
17769 (bmaskdi_vis): Use the `bmask' type.
17770 (bmasksi_vis): Likewise.
17771 * config/sparc/ultra3.md (us3_array): Likewise.
17772 * config/sparc/niagara7.md (n7_array): Likewise.
17773 * config/sparc/niagara4.md (n4_array): Likewise.
17774 * config/sparc/niagara2.md (niag2_vis): Likewise.
17775 (niag3_vis): Likewise.
17776 * config/sparc/niagara.md (niag_vis): Likewise.
17778 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17780 * ipa-comdats.c: Remove optimize check from gate.
17781 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
17782 for functions not optimized.
17783 (ipa_fn_summary_read): Skip optimize check.
17784 (ipa_fn_summary_write): Likewise.
17785 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
17787 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
17789 (can_inline_edge_p): Check flag_pcc_struct_return for match.
17790 (check_callers): Give up on caller which is not optimized.
17791 (inline_small_functions): Likewise.
17792 (ipa_inline): Do not give up when not optimizing.
17793 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
17794 away unoptimizes cdtors.
17795 (whole_program_function_and_variable_visibility): Do
17796 ipa_discover_readonly_nonaddressable_vars in LTO mode.
17797 * ipa.c (process_references): Do not check optimize.
17798 (symbol_table::remove_unreachable_nodes): Update optimize check.
17799 (set_writeonly_bit): Update optimize check.
17800 (pass_ipa_cdtor_merge::gate): Do not check optimize.
17801 (pass_ipa_single_use::gate): Remove.
17803 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17805 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
17806 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
17807 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
17808 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
17809 permute_load, permute_store, adjust_extract, adjust_splat,
17810 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
17811 replace_swap_with_copy, dump_swap_insn_table,
17812 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
17813 recombine_lvx_pattern, recombine_stvx_pattern,
17814 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
17815 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
17816 to file rs6000-p8swap.c.
17817 * config/rs6000/rs6000-p8swap.c: New file.
17818 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17819 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17820 and rs6000*-*-* targets.
17822 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17824 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17826 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17828 * lto-wrapper.c (merge_and_complain): Do not merge
17829 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17830 fsigned_zeros, ftrapping_math, fwrapv.
17831 (append_compiler_options): Do not track these options.
17832 (append_linker_options): Likewie
17834 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17836 * cgraphunit.c (cgraph_node::finalize_function): When
17837 !flag_toplevel_reorde set no_reorder flag.
17838 (varpool_node::finalize_decl): Likewise.
17839 (symbol_table::compile): Drop no toplevel reorder path.
17841 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17843 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17844 edges; zero probability is not better than uninitialized.
17846 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17848 * asan.h (asan_sanitize_allocas_p): Declare.
17849 * asan.c (asan_sanitize_allocas_p): New function.
17850 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17851 (handle_builtin_alloca): Likewise.
17852 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17853 if !asan_sanitize_allocas_p.
17854 * params.def (asan-instrument-allocas): Add new option.
17855 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17856 * opts.c (common_handle_option): Disable allocas sanitization for
17859 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17861 * asan.c: Include gimple-fold.h.
17862 (get_last_alloca_addr): New function.
17863 (handle_builtin_stackrestore): Likewise.
17864 (handle_builtin_alloca): Likewise.
17865 (asan_emit_allocas_unpoison): Likewise.
17866 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17867 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17868 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17869 (instrument_builtin_call): Pass gimple iterator to
17870 get_mem_refs_of_builtin_call.
17871 (last_alloca_addr): New global.
17872 * asan.h (asan_emit_allocas_unpoison): Declare.
17873 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17874 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17875 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17876 if function calls alloca.
17877 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17878 * gimple-fold.h (replace_call_with_value): Declare.
17879 * internal-fn.c: Include asan.h.
17880 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17881 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17883 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17885 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17886 (C_SELFTEST_FLAGS): New.
17887 (CPP_SELFTEST_FLAGS): New.
17888 (SELFTEST_DEPS): New, from deps of s-selftest.
17889 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17890 (CPP_SELFTEST_DEPS): New.
17891 (selftest): Add dependency on s-selftest-c++.
17892 (s-selftest): Rename to...
17893 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17894 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17895 than SELFTEST_FLAGS.
17896 (selftest-gdb): Rename to...
17897 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17899 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17900 (selftest-valgrind): Rename to...
17901 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17903 (selftest-valgrind): Reintroduce as an alias for
17904 selftest-c-valgrind.
17905 (s-selftest-c++): New.
17906 (selftest-c++-gdb): New.
17907 (selftest-c++-valgrind): New.
17909 2017-07-06 Olivier Hainque <hainque@adacore.com>
17911 * gcc.c (process_command): When deciding if undefined variables
17912 should be ignored when processing specs, accept "gcc -v" as well.
17914 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17916 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17917 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17919 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17921 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17922 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17923 * config/arm/arm-tables.opt: Regenerate.
17924 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17926 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17928 2017-07-06 Carl Love <cel@us.ibm.com>
17930 * ChangeLog: Clean up from mid air collision
17932 2017-07-06 Carl Love <cel@us.ibm.com>
17934 * config/rs6000/rs6000-c.c: Add support for built-in functions
17935 vector signed int vec_subc (vector signed int, vector signed int);
17936 vector signed __int128 vec_subc (vector signed __int128,
17937 vector signed __int128);
17938 vector unsigned __int128 vec_subc (vector unsigned __int128,
17939 vector unsigned __int128);
17940 vector signed int vec_sube (vector signed int, vector signed int,
17941 vector signed int);
17942 vector unsigned int vec_sube (vector unsigned int,
17943 vector unsigned int,
17944 vector unsigned int);
17945 vector signed __int128 vec_sube (vector signed __int128,
17946 vector signed __int128,
17947 vector signed__int128);
17948 vector unsigned __int128 vec_sube (vector unsigned __int128,
17949 vector unsigned __int128,
17950 vector unsigned __int128);
17951 vector signed int vec_subec (vector signed int, vector signed int,
17952 vector signed int);
17953 vector unsigned int vec_subec (vector unsigned int,
17954 vector unsigned int,
17955 vector unsigned int);
17956 vector signed __int128 vec_subec (vector signed __int128,
17957 vector signed __int128,
17958 vector signed__int128);
17959 vector unsigned __int128 vec_subec (vector unsigned __int128,
17960 vector unsigned __int128,
17961 vector unsigned __int128);
17962 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17963 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17964 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17965 BU_ALTIVEC_OVERLOAD_X definitions.
17966 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17967 * doc/extend.texi: Update the built-in documentation file for the new
17968 built-in functions.
17970 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17973 * diagnostic-show-locus.c (layout::layout): Use start and finish
17974 spelling location for the start and finish of each range.
17975 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17976 Add unused aspect param.
17977 * input.c (expand_location_1): Add "aspect" param, and use it
17978 to access the correct part of the location.
17979 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17981 (expand_location_to_spelling_point): Likewise.
17982 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17983 param, and pass it to expand_location_1.
17985 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17987 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17988 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17989 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17990 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17991 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17992 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17993 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17994 _mm_maskz_getmant_ss): New intrinsics.
17995 (__builtin_ia32_getexpss128_mask): Changed to ...
17996 __builtin_ia32_getexpss128_round ... this.
17997 (__builtin_ia32_getexpsd128_mask): Changed to ...
17998 __builtin_ia32_getexpsd128_round ... this.
17999 * config/i386/i386-builtin-types.def
18000 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18001 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18002 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18003 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18004 __builtin_ia32_getmantss_mask_round): New builtins.
18005 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18006 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18007 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18008 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18009 * config/i386/sse.md
18010 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18011 avx512f_sgetexp<mode><mask_scalar_name>
18012 <round_saeonly_scalar_name> ... this.
18013 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18014 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18015 vgetexp<ssescalarmodesuffix>
18016 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18017 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18018 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18019 avx512f_vgetmant<mode><mask_scalar_name>
18020 <round_saeonly_scalar_name> ... this.
18021 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18022 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18023 vgetmant<ssescalarmodesuffix>
18024 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18025 %0<mask_scalar_operand4>, %1, %2
18026 <round_saeonly_scalar_mask_op4>, %3} ... this.
18027 * config/i386/subst.md (mask_scalar_operand4,
18028 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18029 round_saeonly_scalar_nimm_predicate): New subst attributes.
18031 2017-07-06 Julia Koval <julia.koval@intel.com>
18033 * config/i386/i386.c (ix86_erase_embedded_rounding):
18034 Remove code for old rounding pattern.
18036 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18038 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18040 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18042 * doc/sourcebuild.texi (Test Directives, Variants of
18043 dg-require-support): Add documentation for dg-require-stack-check.
18045 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18047 * config/i386/subst.md (mask_scalar, round_scalar,
18048 round_saeonly_scalar): New meta-templates.
18049 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18050 round_scalar_mask_operand3, round_scalar_mask_op3,
18051 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18052 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18053 round_saeonly_scalar_constraint,
18054 round_saeonly_scalar_prefix): New subst attribute.
18055 * config/i386/sse.md
18056 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18057 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18058 <round_scalar_name> ... this.
18059 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18060 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18061 <round_scalar_name> ... this.
18062 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18063 <sse>_vm<code><mode>3<mask_scalar_name>
18064 <round_saeonly_scalar_name> ... this.
18065 (v<plusminus_mnemonic><ssescalarmodesuffix>
18066 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18067 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18068 v<plusminus_mnemonic><ssescalarmodesuffix>
18069 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18070 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18071 (v<multdiv_mnemonic><ssescalarmodesuffix>
18072 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18073 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18074 v<multdiv_mnemonic><ssescalarmodesuffix>
18075 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18076 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18077 (v<maxmin_float><ssescalarmodesuffix>
18078 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18079 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18080 v<maxmin_float><ssescalarmodesuffix>
18081 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18082 %0<mask_scalar_operand3>, %1, %<iptr>2
18083 <round_saeonly_scalar_mask_op3>} ... this.
18085 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18087 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18088 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18090 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18091 Alan Hayward <alan.hayward@arm.com>
18092 David Sherwood <david.sherwood@arm.com>
18094 * combine.c (simplify_if_then_else): Remove "enum" before
18096 * compare-elim.c (can_eliminate_compare): Likewise.
18097 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18099 (aarch64_lookup_simd_builtin_type): Likewise.
18100 (aarch64_simd_builtin_type): Likewise.
18101 (aarch64_init_simd_builtin_types): Likewise.
18102 (aarch64_simd_expand_args): Likewise.
18103 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18105 (aarch64_reverse_mask): Likewise.
18106 (aarch64_simd_emit_reg_reg_move): Likewise.
18107 (aarch64_gen_adjusted_ldpstp): Likewise.
18108 (aarch64_ccmp_mode_to_code): Likewise.
18109 (aarch64_operands_ok_for_ldpstp): Likewise.
18110 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18111 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18113 (aarch64_min_divisions_for_recip_mul): Likewise.
18114 (aarch64_reassociation_width): Likewise.
18115 (aarch64_get_condition_code_1): Likewise.
18116 (aarch64_simd_emit_reg_reg_move): Likewise.
18117 (aarch64_simd_attr_length_rglist): Likewise.
18118 (aarch64_reverse_mask): Likewise.
18119 (aarch64_operands_ok_for_ldpstp): Likewise.
18120 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18121 (aarch64_gen_adjusted_ldpstp): Likewise.
18122 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18124 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18125 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18126 (arm_lookup_simd_builtin_type): Likewise.
18127 (arm_simd_builtin_type): Likewise.
18128 (arm_init_simd_builtin_types): Likewise.
18129 (arm_expand_builtin_args): Likewise.
18130 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18131 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18132 (ft32_setup_incoming_varargs): Likewise.
18133 (ft32_function_arg): Likewise.
18134 (ft32_function_arg_advance): Likewise.
18135 (ft32_pass_by_reference): Likewise.
18136 (ft32_arg_partial_bytes): Likewise.
18137 (ft32_valid_pointer_mode): Likewise.
18138 (ft32_addr_space_pointer_mode): Likewise.
18139 (ft32_addr_space_legitimate_address_p): Likewise.
18140 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18142 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18143 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18144 (iamcu_alignment): Likewise.
18145 (canonicalize_vector_int_perm): Likewise.
18146 (ix86_noce_conversion_profitable_p): Likewise.
18147 (ix86_mpx_bound_mode): Likewise.
18148 (ix86_operands_ok_for_move_multiple): Likewise.
18149 * config/microblaze/microblaze-protos.h
18150 (microblaze_expand_conditional_branch_reg): Likewise.
18151 * config/microblaze/microblaze.c
18152 (microblaze_expand_conditional_branch_reg): Likewise.
18153 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18155 (rs6000_reassociation_width): Likewise.
18156 (rs6000_invalid_binary_op): Likewise.
18157 (fusion_p9_p): Likewise.
18158 (emit_fusion_p9_load): Likewise.
18159 (emit_fusion_p9_store): Likewise.
18160 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18162 (riscv_hard_regno_mode_ok_p): Likewise.
18163 (riscv_address_insns): Likewise.
18164 (riscv_split_symbol): Likewise.
18165 (riscv_legitimize_move): Likewise.
18166 (riscv_function_value): Likewise.
18167 (riscv_hard_regno_nregs): Likewise.
18168 (riscv_expand_builtin): Likewise.
18169 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18170 (riscv_build_integer): Likewise.
18171 (riscv_split_integer): Likewise.
18172 (riscv_legitimate_constant_p): Likewise.
18173 (riscv_cannot_force_const_mem): Likewise.
18174 (riscv_regno_mode_ok_for_base_p): Likewise.
18175 (riscv_valid_base_register_p): Likewise.
18176 (riscv_valid_offset_p): Likewise.
18177 (riscv_valid_lo_sum_p): Likewise.
18178 (riscv_classify_address): Likewise.
18179 (riscv_legitimate_address_p): Likewise.
18180 (riscv_address_insns): Likewise.
18181 (riscv_load_store_insns): Likewise.
18182 (riscv_force_binary): Likewise.
18183 (riscv_split_symbol): Likewise.
18184 (riscv_force_address): Likewise.
18185 (riscv_legitimize_address): Likewise.
18186 (riscv_move_integer): Likewise.
18187 (riscv_legitimize_const_move): Likewise.
18188 (riscv_legitimize_move): Likewise.
18189 (riscv_address_cost): Likewise.
18190 (riscv_subword): Likewise.
18191 (riscv_output_move): Likewise.
18192 (riscv_canonicalize_int_order_test): Likewise.
18193 (riscv_emit_int_order_test): Likewise.
18194 (riscv_function_arg_boundary): Likewise.
18195 (riscv_pass_mode_in_fpr_p): Likewise.
18196 (riscv_pass_fpr_single): Likewise.
18197 (riscv_pass_fpr_pair): Likewise.
18198 (riscv_get_arg_info): Likewise.
18199 (riscv_function_arg): Likewise.
18200 (riscv_function_arg_advance): Likewise.
18201 (riscv_arg_partial_bytes): Likewise.
18202 (riscv_function_value): Likewise.
18203 (riscv_pass_by_reference): Likewise.
18204 (riscv_setup_incoming_varargs): Likewise.
18205 (riscv_print_operand): Likewise.
18206 (riscv_elf_select_rtx_section): Likewise.
18207 (riscv_save_restore_reg): Likewise.
18208 (riscv_for_each_saved_reg): Likewise.
18209 (riscv_register_move_cost): Likewise.
18210 (riscv_hard_regno_mode_ok_p): Likewise.
18211 (riscv_hard_regno_nregs): Likewise.
18212 (riscv_class_max_nregs): Likewise.
18213 (riscv_memory_move_cost): Likewise.
18214 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18215 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18216 (rl78_addr_space_address_mode): Likewise.
18217 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18219 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18220 (rs6000_reassociation_width): Likewise.
18221 (rs6000_invalid_binary_op): Likewise.
18222 (fusion_p9_p): Likewise.
18223 (emit_fusion_p9_load): Likewise.
18224 (emit_fusion_p9_store): Likewise.
18225 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18226 (ok_for_simple_move_operands): Likewise.
18227 (ok_for_simple_move_strict_operands): Likewise.
18228 (ok_for_simple_arith_logic_operands): Likewise.
18229 (visium_legitimize_reload_address): Likewise.
18230 (visium_select_cc_mode): Likewise.
18231 (output_cbranch): Likewise.
18232 (visium_split_double_move): Likewise.
18233 (visium_expand_copysign): Likewise.
18234 (visium_expand_int_cstore): Likewise.
18235 (visium_expand_fp_cstore): Likewise.
18236 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18237 (visium_function_arg): Likewise.
18238 (visium_function_arg_advance): Likewise.
18239 (visium_libcall_value): Likewise.
18240 (visium_setup_incoming_varargs): Likewise.
18241 (visium_legitimate_constant_p): Likewise.
18242 (visium_legitimate_address_p): Likewise.
18243 (visium_legitimize_address): Likewise.
18244 (visium_secondary_reload): Likewise.
18245 (visium_register_move_cost): Likewise.
18246 (visium_memory_move_cost): Likewise.
18247 (prepare_move_operands): Likewise.
18248 (ok_for_simple_move_operands): Likewise.
18249 (ok_for_simple_move_strict_operands): Likewise.
18250 (ok_for_simple_arith_logic_operands): Likewise.
18251 (visium_function_value_1): Likewise.
18252 (rtx_ok_for_offset_p): Likewise.
18253 (visium_legitimize_reload_address): Likewise.
18254 (visium_split_double_move): Likewise.
18255 (visium_expand_copysign): Likewise.
18256 (visium_expand_int_cstore): Likewise.
18257 (visium_expand_fp_cstore): Likewise.
18258 (visium_split_cstore): Likewise.
18259 (visium_select_cc_mode): Likewise.
18260 (visium_split_cbranch): Likewise.
18261 (output_cbranch): Likewise.
18262 (visium_print_operand_address): Likewise.
18263 * expmed.c (flip_storage_order): Likewise.
18264 * expmed.h (emit_cstore): Likewise.
18265 (flip_storage_order): Likewise.
18266 * genrecog.c (validate_pattern): Likewise.
18267 * hsa-gen.c (gen_hsa_addr): Likewise.
18268 * internal-fn.c (expand_arith_overflow): Likewise.
18269 * ira-color.c (allocno_copy_cost_saving): Likewise.
18270 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18271 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18272 (process_invariant_for_inheritance): Likewise.
18273 * lra-eliminations.c (move_plus_up): Likewise.
18274 * omp-low.c (lower_oacc_reductions): Likewise.
18275 * simplify-rtx.c (simplify_subreg): Likewise.
18276 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18277 (TARGET_CHKP_BOUND_MODE): Likewise..
18278 * targhooks.c (default_chkp_bound_mode): Likewise.
18279 (default_setup_incoming_vararg_bounds): Likewise.
18280 * targhooks.h (default_chkp_bound_mode): Likewise.
18281 (default_setup_incoming_vararg_bounds): Likewise.
18282 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18283 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18284 (have_whole_vector_shift): Likewise.
18285 * tree-vect-stmts.c (vectorizable_load): Likewise.
18286 * doc/tm.texi: Regenerate.
18288 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18290 Graceful degrade if Binutils PR21472 is not available.
18293 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18294 .rodata in flash test fails.
18295 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18296 * confgure: Regenerate.
18297 * config.in: Regenerate.
18298 * config/avr/avr.c (avr_asm_named_section)
18299 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18300 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18301 (avr_asm_init_sections): Same.
18303 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18305 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18306 (fma<VH:mode>4_intrinsic): Likewise.
18307 (*fmsub<VCVTF:mode>4): Likewise.
18308 (*fmsub<VH:mode>4_intrinsic): Likewise.
18310 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18313 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18314 Don't depend on "optimize > 0".
18315 (out_movhi_r_mr, out_movqi_mr_r): Same.
18316 (out_movhi_mr_r, out_movqi_r_mr): Same.
18317 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18318 io_address_operand on "optimize > 0".
18320 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18322 * tree-loop-distribution.c: Add general explanantion on the pass.
18323 (generate_loops_for_partition): Mark distributed loop.
18324 (pg_add_dependence_edges): New parameter. Handle alias data
18325 dependence specially and record it in the parameter if asked.
18326 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18327 (init_partition_graph_vertices, add_partition_graph_edge): New.
18328 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18329 (free_partition_graph_vdata, build_partition_graph): New.
18330 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18331 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18332 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18333 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18334 (version_for_distribution_p, finalize_partitions): New.
18335 (distribute_loop): Handle alias data dependence specially. Factor
18336 out loop fusion code as functions and call these functions.
18338 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18340 * tree-loop-distribution.c (classify_partition): New parameter and
18341 better handle reduction statement.
18342 (rdg_build_partitions): Revise comment.
18343 (distribute_loop): Compute statements in all partitions and pass it
18344 to classify_partition.
18346 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18348 * tree-loop-distribution.c (enum partition_type): New.
18349 (struct partition): New field type.
18350 (partition_merge_into): Add parameter. Update partition type.
18351 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18352 (build_rdg_partition_for_vertex): Compute partition type.
18353 (rdg_build_partitions): Dump partition type.
18354 (distribute_loop): Update calls to partition_merge_into.
18356 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18358 * tree-loop-distribution.c (struct ddr_hasher): New.
18359 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18361 (classify_partition): Call get_data_dependence.
18362 (pg_add_dependence_edges): Ditto.
18363 (distribute_loop): Release data dependence hash table.
18365 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18367 * tree-loop-distribution.c (ref_base_address): Delete.
18368 (similar_memory_accesses): Rename ...
18369 (share_memory_accesses): ... to this. Check if partitions access
18370 the same memory reference.
18371 (distribute_loop): Call share_memory_accesses.
18373 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18375 * tree-loop-distribution.c (struct partition): New field recording
18376 its data reference.
18377 (partition_alloc, partition_free): Init and release data refs.
18378 (partition_merge_into): Merge data refs.
18379 (build_rdg_partition_for_vertex): Collect data refs for partition.
18380 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18382 (distribute_loop): Remve data refs from vertice data of partition
18385 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18387 * tree-loop-distribution.c (params.h): Include header file.
18388 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18389 (datarefs_vec): New global var.
18390 (create_rdg_vertices): Use datarefs_vec directly.
18391 (free_rdg): Don't free data references.
18392 (build_rdg): Update use. Don't free data references.
18393 (distribute_loop): Compute global variable for data references.
18394 Bail out if there are too many data references.
18396 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18398 * tree-loop-distribution.c (loop_nest): New global var.
18399 (build_rdg): Use loop directly, rather than loop nest.
18400 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18402 (distribute_loop): Compute global variable loop nest. Update use.
18404 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18406 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18407 (partition_merge_into): New parameter. Dump reason for fusion.
18408 (distribute_loop): Update use of partition_merge_into.
18410 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18412 * tree-loop-distribution.c (bb_top_order_index): New.
18413 (bb_top_order_index_size, bb_top_order_cmp): New.
18414 (stmts_from_loop): Use topological order.
18415 (pass_loop_distribution::execute): Compute and release topological
18416 order for basic blocks.
18418 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18420 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18423 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18425 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18426 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18427 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18428 * internal-fn.def (LOOP_DIST_ALIAS): New.
18429 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18430 (fold_loop_internal_call): ... this.
18431 (vect_loop_dist_alias_call): New function.
18432 (set_uid_loop_bbs): Call fold_loop_internal_call.
18433 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18436 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18439 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18440 Require dead FLAGS_REG at the beginning of a peephole.
18442 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18445 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18446 arguments in the call to __builtin_ia32_sbb_u32.
18447 (_subborrow_u64): Swap _X and _Y arguments in the call to
18448 __builtin_ia32_sbb_u64.
18450 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18453 * tree-vrp.c (compare_assert_loc): Turn into a function template
18454 with stable template parameter. Only test if a->e is NULL,
18455 !a->e == !b->e has been verified already. Use e == NULL or
18456 e != NULL instead of e or ! e tests. If stable is true, don't use
18457 iterative_hash_expr, on the other side allow a or b or both NULL
18458 and sort the NULLs last.
18459 (process_assert_insertions): Sort using compare_assert_loc<false>
18460 instead of compare_assert_loc, later sort using
18461 compare_assert_loc<true> before calling process_assert_insertions_for
18462 in a loop. Use break instead of continue once seen NULL pointer.
18464 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18466 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18467 Cortex-R7 and Cortex-R8 processors.
18469 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18471 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18472 uninitialized while src is not.
18474 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18476 * common/config/arm/arm-common.c: Adjust include path for
18478 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18479 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18480 (arm-cpu-data.h): Likewise.
18481 (arm-cpu-cdata.h): Likewise.
18482 * config/arm/arm-cpu.h: Delete.
18483 * config/arm/arm-cpu-cdata.h: Delete.
18484 * config/arm/arm-cpu-data.h: Delete.
18486 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18488 * config/arm/arm-cpus.in (cortex-a55): New.
18489 (cortex-a75): Likewise.
18490 (cortex-a75.cortex-a55): Likewise.
18491 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18493 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18494 * config/arm/arm-cpu-cdata.h: Regenerate.
18495 * config/arm/arm-cpu-data.h: Regenerate.
18496 * config/arm/arm-cpu.h: Regenerate.
18497 * config/arm/arm-tables.opt: Regenerate.
18498 * config/arm/arm-tune.md: Regenerate.
18500 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18502 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18504 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18506 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18509 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18511 PR tree-optimization/81292
18512 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18513 full_string_p, also call adjust_related_strinfos if the adjustment
18514 is simple, otherwise invalidate related strinfos.
18516 2017-07-04 Martin Liska <mliska@suse.cz>
18519 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18520 newly created variable as DECL_IGNORED_P.
18522 2017-07-04 Martin Liska <mliska@suse.cz>
18525 * ipa-inline.c (inline_small_functions):
18526 Use xstrdup_for_dump.
18528 2017-07-04 Tom de Vries <tom@codesourcery.com>
18530 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18532 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18535 * config/darwin.c (darwin_function_switched_text_sections):
18538 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18540 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18542 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18544 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18546 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18548 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18549 min_profitable_iters, and th as inclusive lower bounds.
18550 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18551 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18552 for min_profitable_iters and min_profitable_estimate.
18553 (vect_transform_loop): Treat th as an inclusive lower bound.
18554 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
18556 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18559 * config/darwin.c (darwin_function_switched_text_sections):
18560 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
18561 in two pieces, and suppress the use of buf.
18563 2017-07-03 Nathan Sidwell <nathan@acm.org>
18565 * hash-table.h (hash_table_mod1): Fix indentation.
18567 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18569 PR middle-end/81290
18570 * predict.c (force_edge_cold): Be more careful about propagation
18572 * profile-count.h (profile_probability::guessed,
18573 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
18575 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
18577 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
18579 * doc/invoke.texi (rcpc architecture extension): Document it.
18581 2017-07-03 Richard Biener <rguenther@suse.de>
18583 PR tree-optimization/60510
18584 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
18585 the scalar reduction PHI and use it.
18586 (vectorizable_reduction): Properly guard the single_defuse_cycle
18587 path for non-SLP reduction chains where we cannot use it.
18588 Rework reduc_def/index and vector type deduction. Rework
18589 vector operand gathering during reduction op code-gen.
18590 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
18591 chains dissolve the chain and leave it to non-SLP reduction
18594 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18596 * tree-data-ref.h (dr_alignment): Declare.
18597 * tree-data-ref.c (dr_alignment): New function.
18598 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
18599 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18601 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
18603 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18605 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
18606 and base_misalignment fields.
18607 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
18608 * tree-data-ref.c: Include builtins.h.
18609 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
18610 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
18611 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
18612 * tree-vect-data-refs.c: Include tree-cfg.h.
18613 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
18614 fields instead of calculating an alignment here.
18615 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
18616 innermost_loop_behavior fields.
18618 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18620 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
18622 (DR_STEP_ALIGNMENT): New macro.
18623 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
18624 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
18625 (create_data_ref): Print it.
18626 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
18627 to tell whether the step preserves vector (mis)alignment.
18628 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18629 Move the check for an integer step and generalise to all INTEGER_CST.
18630 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
18631 Print the outer step alignment.
18633 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18635 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
18636 with offset_alignment.
18637 (DR_ALIGNED_TO): Delete.
18638 (DR_OFFSET_ALIGNMENT): New macro.
18639 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
18640 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
18641 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
18642 (create_data_ref): Likewise.
18643 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18644 (vect_analyze_data_refs): Likewise.
18645 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
18646 creating dummy innermost behavior.
18648 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18650 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
18651 with a "innermost_loop_behavior *" and refeence tree.
18652 * tree-data-ref.c (dr_analyze_innermost): Likewise.
18653 (create_data_ref): Update call accordingly.
18654 * tree-predcom.c (find_looparound_phi): Likewise.
18656 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18658 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
18659 fields with dr_wrt_vec_loop.
18660 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
18661 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
18662 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
18663 (vect_dr_behavior): New function.
18664 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18665 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
18666 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
18667 track whether the step preserves the misalignment.
18668 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18669 Use vect_dr_behavior.
18670 (vect_setup_realignment): Update call accordingly.
18671 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
18672 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
18673 call to vect_create_addr_base_for_vector_ref.
18674 (vect_create_cond_for_align_checks): Likewise.
18675 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
18676 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
18677 (vect_recog_mask_conversion_pattern): Likewise.
18678 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
18679 (new_stmt_vec_info): Remove redundant zeroing.
18681 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18683 * common/config/arm/arm-common.c (arm_be8_option): New function.
18684 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
18685 (ISA_ARMv6): Add isa_bit_be8.
18686 * config/arm/arm.h (arm_be8_option): Add prototype.
18687 (BE8_SPEC_FUNCTION): New define.
18688 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
18689 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
18690 (mlittle-endian): Similarly.
18691 (mbe8, mbe32): New options.
18692 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
18693 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
18695 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18697 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
18699 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18701 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
18702 (cleanup_tree_cfg_bb): Use it.
18703 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
18705 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
18707 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18710 * loop-doloop.c (add_test): Update profile.
18712 2017-07-03 Martin Liska <mliska@suse.cz>
18715 * sanopt.c (rewrite_usage_of_param): New function.
18716 (sanitize_rewrite_addressable_params): Likewise.
18717 (pass_sanopt::execute): Call rewrite_usage_of_param.
18719 2017-07-03 Richard Biener <rguenther@suse.de>
18721 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
18722 back to using VIEW_CONVERT_EXPR.
18724 2017-07-03 Martin Liska <mliska@suse.cz>
18727 * doc/extend.texi: Document when a resolver function is
18728 generated for target_clones.
18730 2017-07-03 Martin Liska <mliska@suse.cz>
18732 * asan.c (asan_emit_stack_protection): Unpoison just red zones
18733 and shadow memory of auto variables which are subject of
18734 use-after-scope sanitization.
18735 (asan_expand_mark_ifn): Add do set only when is_poison.
18737 2016-07-03 Richard Biener <rguenther@suse.de>
18739 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18741 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18742 (vectorizable_reduction): Perform reduction PHI creation when
18743 visiting a reduction PHI and adjust and simplify code generation
18744 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18745 (vect_transform_loop): Visit reduction PHIs.
18746 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18747 defs into the SLP tree.
18748 (vect_build_slp_tree): Reduction defs terminate the recursion.
18749 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18751 (vect_get_vec_defs_for_stmt_copy): Export.
18752 (vect_get_vec_defs): Likewise.
18753 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18755 (vect_get_vec_defs_for_stmt_copy): Declare.
18756 (vect_get_vec_defs): Likewise.
18758 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18760 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18761 parameter with a "loop" parameter and use it instead of the
18762 loop containing DR_STMT. Don't check simple_iv when doing
18763 BB analysis. Describe the two analysis modes in the comment.
18765 2017-07-03 Tom de Vries <tom@codesourcery.com>
18767 PR tree-optimization/69468
18768 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
18769 (find_same_succ_bb): Handle ignore_edge_flags.
18771 2017-07-03 Tom de Vries <tom@codesourcery.com>
18773 PR tree-optimization/81192
18774 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
18776 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
18778 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
18780 2017-07-03 Tom de Vries <tom@codesourcery.com>
18782 PR tree-optimization/81192
18783 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
18784 BB_SAME_SUCC (bb) == NULL.
18786 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18788 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
18791 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18793 * dumpfile.c: Include profile-count.h
18794 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
18796 (insert_cond_bb): Update profile.
18797 * tree-cfg.h (insert_cond_bb): Update prototype.
18798 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
18799 * tree-dump.c: Do not include tree-cfg.
18801 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18803 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
18805 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18807 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
18810 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18812 * tree-complex.c (expand_complex_div_wide): update profile.
18814 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18815 Alan Hayward <alan.hayward@arm.com>
18816 David Sherwood <david.sherwood@arm.com>
18818 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18819 (CORETYPES_H): New define.
18820 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18821 (insn-modes-inline.h, s-modes-inline-h): New rules.
18822 (generated_files): Add insn-modes-inline.h.
18823 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18824 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18825 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18826 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18827 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18828 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18829 (build/gencodes.o, build/genconditions.o): Likewise.
18830 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18831 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18832 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18833 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18834 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18835 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18836 * coretypes.h: Include everything up to real.h for generators.
18837 Include insn-modes.h first. Include wide-int-print.h after
18838 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18839 * machmode.h: Don't include insn-modes.h here.
18840 * function-tests.c: Remove includes of signop.h, machmode.h,
18841 double-int.h and wide-int.h.
18843 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18845 * optc-save-gen.awk: Likewise.
18846 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18847 * godump.c: Remove include of wide-int-print.h.
18848 * pretty-print.h: Likewise.
18849 * wide-int-print.cc: Likewise.
18850 * wide-int.cc: Likewise.
18851 * hash-map-tests.c: Remove include of signop.h.
18852 * hash-set-tests.c: Likewise.
18853 * rtl-tests.c: Likewise.
18854 * mkconfig.sh: Remove include of machmode.h.
18855 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18857 (emit_insn_modes_inline_h): ...this new function. Emit the code
18858 into an insn-modes-inline.h header file, adding appropriate
18859 include guards and end comments.
18860 (emit_insn_modes_c_header): Remove include of machmode.h.
18861 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18863 (main): Handle -i flag and call emit_insn_modes_inline_h when
18866 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18868 * tree-ssa-strlen.c (strinfo): Rename the length field to
18869 nonzero_chars. Add a full_string_p field.
18870 (compare_nonzero_chars, zero_length_string_p): New functions.
18871 (get_addr_stridx): Add an offset_out parameter.
18872 Use compare_nonzero_chars.
18873 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18874 (new_strinfo): Update after above changes to strinfo.
18875 (set_endptr_and_length): Set full_string_p.
18876 (get_string_length): Update after above changes to strinfo.
18877 (unshare_strinfo): Update call to new_strinfo.
18878 (maybe_invalidate): Likewise.
18879 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18880 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18881 as a uhwi instead of an shwi. Update after above changes to
18882 strinfo and new_strinfo.
18883 (zero_length_string): Assert that chainsi contains full strings.
18884 Use zero_length_string_p. Update call to new_strinfo.
18885 (adjust_related_strinfos): Update after above changes to strinfo.
18886 Copy full_string_p from origsi.
18887 (adjust_last_stmt): Use zero_length_string_p.
18888 (handle_builtin_strlen): Update after above changes to strinfo and
18889 new_strinfo. Install the lhs as the string length if the previous
18890 entry didn't describe a full string.
18891 (handle_builtin_strchr): Update after above changes to strinfo
18893 (handle_builtin_strcpy): Likewise.
18894 (handle_builtin_strcat): Likewise.
18895 (handle_builtin_malloc): Likewise.
18896 (handle_pointer_plus): Likewise.
18897 (handle_builtin_memcpy): Likewise. Track nonzero characters
18898 that aren't necessarily followed by a nul terminator.
18899 (handle_char_store): Likewise.
18901 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18903 PR tree-optimization/80769
18904 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18905 for malloc and calloc. Document the new invariant that all related
18906 strinfos have delayed lengths or none do.
18907 (verify_related_strinfos): Move earlier in file.
18908 (set_endptr_and_length): New function, split out from...
18909 (get_string_length): ...here. Also set the lengths of related
18911 (zero_length_string): Assert that chainsi has known (rather than
18913 (adjust_related_strinfos): Likewise.
18915 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18917 PR tree-optimization/81136
18918 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18919 assert that two references with the same misalignment have the same
18920 compile-time misalignment if those compile-time misalignments
18923 2017-07-01 Andi Kleen <ak@linux.intel.com>
18925 * print-tree.c (print_node): Print all attributes.
18927 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18929 * cfg.c (scale_bbs_frequencies): New function.
18930 * cfg.h (scale_bbs_frequencies): Declare it.
18931 * cfgloopanal.c (single_likely_exit): Cleanup.
18932 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18934 (scale_loop_profile): Likewise.
18935 (loop_version): Likewise.
18936 (create_empty_loop_on_edge): Update.
18937 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18938 scale_loop_frequencies, scale_loop_profile, loopify,
18939 loop_version): Update prototypes.
18940 * modulo-sched.c (sms_schedule): Update.
18941 * predict.c (unlikely_executed_edge_p): Also check probability.
18942 (probably_never_executed_edge_p): Fix typo.
18943 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18944 * tree-parloops.c (gen_parallel_loop): Update.
18945 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18946 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18947 * tree-ssa-loop-split.c (split_loop): Update.
18948 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18949 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18950 (vect_loop_versioning): Update.
18951 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18953 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18955 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18957 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18959 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18960 to keep profile consistent.
18962 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18964 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18965 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18966 * profile-count.h (max_safe_multiplier): Make unsigned.
18967 (profile_count::guessed_zero): New.
18969 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18971 * bb-reorder.c (fix_up_crossing_landing_pad,
18972 fix_crossing_conditional_branches): Use make_single_succ_edge
18973 to keep profile consistent.
18975 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18977 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18980 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18983 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18984 the right scopes, make sure cond_jump isn't preserved between multiple
18985 iterations. Search for fallthru edge whenever there are 3+ edges and
18986 use find_fallthru_edge for it.
18988 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18990 Patch by Alexander Monakov <amonakov@ispras.ru>
18991 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18992 probabilities consistently.
18994 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18996 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18997 emit_cmp_and_jump_insns.
18999 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19002 * tree-inline.c (expand_call_inline): Combine profile statuses.
19004 2017-06-30 Andrew Pinski <apinski@cavium.com>
19006 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19007 fold_stmt returned true.
19009 2017-06-30 Nathan Sidwell <nathan@acm.org>
19011 * ggc.h (empty_string): Delete.
19012 * cfgexpand.c (expand_asm_stmt): Use plain "".
19013 * optabs.c (expand_asm_memory_barrier): Likewise.
19014 * stringpool.c (empty_string): Delete.
19015 (digit_vector, digit_string): Delete.
19016 (ggc_alloc_string): Use plain "", don't optimize single digit
19017 strings. Use ggc_alloc_atomic.
19019 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19021 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19022 comparison set and one other set, use the cost of the non-comparison
19025 2017-06-30 Nathan Sidwell <nathan@acm.org>
19027 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19030 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19032 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19033 loops. Remove now unneeded calls to gimple_switch_set_label() that
19034 just set removed labels to NULL_TREE.
19036 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19038 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19039 (set_range_info): ...here. Only call set_range_info_raw if domain
19041 (set_nonzero_bits): Call set_range_info_raw.
19042 * tree-ssanames.h (set_range_info_raw): New.
19044 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19047 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19048 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19049 of nonimmediate_operand and <store_mask_constraint> instead of m
19050 for the input operand. For V8FI iterator, always split if input
19051 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19052 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19053 <store_mask_predicate> instead of register_operand and
19054 <store_mask_constraint> instead of v for the input operand. Make
19055 sure both operands aren't MEMs for if not <mask_applied>.
19057 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19059 * lto-wrapper.c (copy_file) Close both file descriptors before
19062 2017-06-30 Martin Liska <mliska@suse.cz>
19065 * multiple_target.c (create_dispatcher_calls): Make ifunc
19066 also for function that don't have calls or are not referenced.
19068 2017-06-30 Richard Biener <rguenther@suse.de>
19070 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19071 analyze the first scalar stmt. Move vector type computation
19072 for the BB case here from ...
19073 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19074 live operation processing in the SLP case properly.
19076 2017-06-30 Richard Biener <rguenther@suse.de>
19078 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19080 2017-06-30 Martin Liska <mliska@suse.cz>
19083 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19084 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19086 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19088 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19091 2017-06-30 Richard Biener <rguenther@suse.de>
19093 PR tree-optimization/81249
19094 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19095 condition reduction result to original scalar type.
19097 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19099 * profile-count.h (enum profile_quality): Fix typos and whitespace
19102 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19104 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19105 type for branch probabilities.
19107 2017-06-29 Julian Brown <julian@codesourcery.com>
19108 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19110 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19111 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19112 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19113 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19115 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19117 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19118 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19119 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19120 CC usage from generic code to here.
19121 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19122 CC usage into the target macros.
19124 2017-06-29 Maya Rashish <coypu@sdf.org>
19126 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19129 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19131 * arm/arm-builtins.c: Include profile-count.h
19132 * except.c (sjlj_emit_function_enter): Use
19133 profile_probability::unlikely.
19135 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19137 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19138 and tocrel_offset be pointer args rather than implicitly using
19140 (legitimate_constant_pool_address_p, rs6000_emit_move,
19141 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19142 tocrel_offset and use in toc_relative_expr_p call.
19143 (print_operand, print_operand_address): Use static tocrel_base_oac
19144 and tocrel_offset_oac.
19145 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19148 2017-06-29 Maya Rashish <coypu@sdf.org>
19150 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19152 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19154 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19155 objects, take into account only the alignment of 'op0' and 'mode1' if
19158 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19160 * ccmp.c (ccmp_tree_comparison_p): New function.
19161 (ccmp_candidate_p): Update to use above function.
19162 (get_compare_parts): New function.
19163 (expand_ccmp_next): Update to use new functions.
19164 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19166 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19167 take mode as argument.
19168 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19169 * expr.c (expand_expr_real_1): Pass mode as argument.
19171 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19173 * combine.c (combine_instructions): Print insns to dump_file, together
19176 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19178 * asan.c (asan_emit_stack_protection): Update.
19179 (create_cond_insert_point): Update.
19180 * auto-profile.c (afdo_propagate_circuit): Update.
19181 * basic-block.h (struct edge_def): Turn probability to
19182 profile_probability.
19183 (EDGE_FREQUENCY): Update.
19184 * bb-reorder.c (find_traces_1_round): Update.
19185 (better_edge_p): Update.
19186 (sanitize_hot_paths): Update.
19187 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19188 (make_single_succ_edge): Update.
19189 (check_bb_profile): Update.
19190 (dump_edge_info): Update.
19191 (update_bb_profile_for_threading): Update.
19192 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19193 probabilitycount to 0.
19194 * cfgbuild.c (compute_outgoing_frequencies): Update.
19195 * cfgcleanup.c (try_forward_edges): Update.
19196 (outgoing_edges_match): Update.
19197 (try_crossjump_to_edge): Update.
19198 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19199 (expand_gimple_tailcall): Update.
19200 (construct_init_block): Use make_single_succ_edge.
19201 (construct_exit_block): Use make_single_succ_edge.
19202 * cfghooks.c (verify_flow_info): Update.
19203 (redirect_edge_succ_nodup): Update.
19204 (split_edge): Update.
19205 (account_profile_record): Update.
19206 * cfgloopanal.c (single_likely_exit): Update.
19207 * cfgloopmanip.c (scale_loop_profile): Update.
19208 (set_zero_probability): Remove.
19209 (duplicate_loop_to_header_edge): Update.
19210 * cfgloopmanip.h (loop_version): Update prototype.
19211 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19212 (force_nonfallthru_and_redirect): Update.
19213 (update_br_prob_note): Update.
19214 (rtl_verify_edges): Update.
19215 (purge_dead_edges): Update.
19216 (rtl_lv_add_condition_to_bb): Update.
19217 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19218 * cgraphunit.c (init_lowered_empty_function): Update.
19219 (cgraph_node::expand_thunk): Update.
19220 * cilk-common.c: Include profile-count.h
19221 * dojump.c (inv): Remove.
19222 (jumpifnot): Update.
19223 (jumpifnot_1): Update.
19224 (do_jump_1): Update.
19226 (do_jump_by_parts_greater_rtx): Update.
19227 (do_compare_rtx_and_jump): Update.
19228 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19229 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19230 * dwarf2cfi.c: Include profile-count.h
19231 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19232 (sjlj_emit_dispatch_table): Likewise.
19233 * explow.c: Include profile-count.h
19234 * expmed.c (emit_store_flag_force): Update.
19235 (do_cmp_and_jump): Update.
19236 * expr.c (compare_by_pieces_d::generate): Update.
19237 (compare_by_pieces_d::finish_mode): Update.
19238 (emit_block_move_via_loop): Update.
19239 (store_expr_with_bounds): Update.
19240 (store_constructor): Update.
19241 (expand_expr_real_2): Update.
19242 (expand_expr_real_1): Update.
19243 * expr.h (try_casesi, try_tablejump): Update prototypes.
19244 * gimple-pretty-print.c (dump_probability): Update.
19245 (dump_profile): New.
19246 (dump_gimple_label): Update.
19247 (dump_gimple_bb_header): Update.
19248 * graph.c (draw_cfg_node_succ_edges): Update.
19249 * hsa-gen.c (convert_switch_statements): Update.
19250 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19251 (find_if_case_1): Update.
19252 (find_if_case_2): Update.
19253 * internal-fn.c (expand_arith_overflow_result_store): Update.
19254 (expand_addsub_overflow): Update.
19255 (expand_neg_overflow): Update.
19256 (expand_mul_overflow): Update.
19257 (expand_vector_ubsan_overflow): Update.
19258 * ipa-cp.c (good_cloning_opportunity_p): Update.
19259 * ipa-split.c (split_function): Use make_single_succ_edge.
19260 * ipa-utils.c (ipa_merge_profiles): Update.
19261 * loop-doloop.c (add_test): Update.
19262 (doloop_modify): Update.
19263 * loop-unroll.c (compare_and_jump_seq): Update.
19264 (unroll_loop_runtime_iterations): Update.
19265 * lra-constraints.c (lra_inheritance): Update.
19266 * lto-streamer-in.c (input_cfg): Update.
19267 * lto-streamer-out.c (output_cfg): Update.
19268 * mcf.c (adjust_cfg_counts): Update.
19269 * modulo-sched.c (sms_schedule): Update.
19270 * omp-expand.c (expand_omp_for_init_counts): Update.
19271 (extract_omp_for_update_vars): Update.
19272 (expand_omp_ordered_sink): Update.
19273 (expand_omp_for_ordered_loops): Update.
19274 (expand_omp_for_generic): Update.
19275 (expand_omp_for_static_nochunk): Update.
19276 (expand_omp_for_static_chunk): Update.
19277 (expand_cilk_for): Update.
19278 (expand_omp_simd): Update.
19279 (expand_omp_taskloop_for_outer): Update.
19280 (expand_omp_taskloop_for_inner): Update.
19281 * omp-simd-clone.c (simd_clone_adjust): Update.
19282 * optabs.c (expand_doubleword_shift): Update.
19283 (expand_abs): Update.
19284 (emit_cmp_and_jump_insn_1): Update.
19285 (expand_compare_and_swap_loop): Update.
19286 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19287 * predict.c (predictable_edge_p): Update.
19288 (edge_probability_reliable_p): Update.
19289 (set_even_probabilities): Update.
19290 (combine_predictions_for_insn): Update.
19291 (combine_predictions_for_bb): Update.
19292 (propagate_freq): Update.
19293 (estimate_bb_frequencies): Update.
19294 (force_edge_cold): Update.
19295 * profile-count.c (profile_count::dump): Add missing space into dump.
19296 (profile_count::debug): Add newline.
19297 (profile_count::differs_from_p): Explicitly convert to unsigned.
19298 (profile_count::stream_in): Update.
19299 (profile_probability::dump): New member function.
19300 (profile_probability::debug): New member function.
19301 (profile_probability::differs_from_p): New member function.
19302 (profile_probability::differs_lot_from_p): New member function.
19303 (profile_probability::stream_in): New member function.
19304 (profile_probability::stream_out): New member function.
19305 * profile-count.h (profile_count_quality): Rename to ...
19306 (profile_quality): ... this one.
19307 (profile_probability): New.
19308 (profile_count): Update.
19309 * profile.c (compute_branch_probabilities): Update.
19310 * recog.c (peep2_attempt): Update.
19311 * sched-ebb.c (schedule_ebbs): Update.
19312 * sched-rgn.c (find_single_block_region): Update.
19313 (compute_dom_prob_ps): Update.
19314 (schedule_region): Update.
19315 * sel-sched-ir.c (compute_succs_info): Update.
19316 * stmt.c (struct case_node): Update.
19317 (do_jump_if_equal): Update.
19318 (get_outgoing_edge_probs): Update.
19319 (conditional_probability): Update.
19320 (emit_case_dispatch_table): Update.
19321 (expand_case): Update.
19322 (expand_sjlj_dispatch_table): Update.
19323 (emit_case_nodes): Update.
19324 * targhooks.c: Update.
19325 * tracer.c (better_p): Update.
19326 (find_best_successor): Update.
19327 * trans-mem.c (expand_transaction): Update.
19328 * tree-call-cdce.c: Update.
19329 * tree-cfg.c (gimple_split_edge): Upate.
19330 (move_sese_region_to_fn): Upate.
19331 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19332 * tree-eh.c (lower_resx): Upate.
19333 (cleanup_empty_eh_move_lp): Upate.
19334 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19335 * tree-inline.c (copy_edges_for_bb): Update.
19336 (copy_cfg_body): Update.
19337 * tree-parloops.c (gen_parallel_loop): Update.
19338 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19339 (gimple_gen_time_profiler): Update.
19340 * tree-ssa-dce.c (remove_dead_stmt): Update.
19341 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19342 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19343 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19344 (unloop_loops): Update.
19345 (try_peel_loop): Update.
19346 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19347 * tree-ssa-loop-split.c (connect_loops): Update.
19348 (split_loop): Update.
19349 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19350 (hoist_guard): Update.
19351 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19352 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19353 (value_replacement): Update.
19354 * tree-ssa-reassoc.c (branch_fixup): Update.
19355 * tree-ssa-tail-merge.c (replace_block_by): Update.
19356 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19357 (create_edge_and_update_destination_phis): Update.
19358 (compute_path_counts): Update.
19359 (recompute_probabilities): Update.
19360 (update_joiner_offpath_counts): Update.
19361 (freqs_to_counts_path): Update.
19362 (duplicate_thread_path): Update.
19363 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19364 (struct switch_conv_info): Update.
19365 (gen_inbound_check): Update.
19366 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19367 (vect_do_peeling): Update.
19368 (vect_loop_versioning): Update.
19369 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19370 (optimize_mask_stores): Update.
19371 * ubsan.c (ubsan_expand_null_ifn): Update.
19372 * value-prof.c (gimple_divmod_fixed_value): Update.
19373 (gimple_divmod_fixed_value_transform): Update.
19374 (gimple_mod_pow2): Update.
19375 (gimple_mod_pow2_value_transform): Update.
19376 (gimple_mod_subtract): Update.
19377 (gimple_mod_subtract_transform): Update.
19378 (gimple_ic): Update.
19379 (gimple_stringop_fixed_value): Update.
19380 (gimple_stringops_transform): Update.
19381 * value-prof.h: Update.
19383 2017-06-29 Carl Love <cel@us.ibm.com>
19385 * config/rs6000/rs6000-c.c: Add support for built-in functions
19386 vector signed int vec_signed (vector float);
19387 vector signed long long vec_signed (vector double);
19388 vector signed int vec_signed2 (vector double, vector double);
19389 vector signed int vec_signede (vector double);
19390 vector signed int vec_signedo (vector double);
19391 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19392 instruction generator.
19393 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19394 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19395 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19397 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19398 vunsignede_v2df): Add define_expands.
19399 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19400 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19401 VEC_UNSIGNEDO): Add definitions.
19402 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19403 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19404 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19405 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19406 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19407 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19408 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19409 vec_unsignede, vec_unsignedo): Add builtin defines.
19410 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19412 * doc/extend.texi: Update the built-in documentation file for the
19413 new built-in functions.
19415 2017-06-29 Richard Biener <rguenther@suse.de>
19417 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19418 reduction chains to LOOP_VINFO_REDUCTIONS.
19419 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19420 SLP reductions after processing reduction chains.
19422 2017-06-29 Nathan Sidwell <nathan@acm.org>
19424 * builtins.c (fold_builtin_FUNCTION): Use
19425 lang_hooks.decl_printable_name.
19427 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19429 PR middle-end/81194
19430 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19431 with only one label.
19432 * stmt.c (expand_case): Assert NCASES is greater than one.
19434 2017-06-29 Richard Biener <rguenther@suse.de>
19436 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19438 (group_case_labels): Likewise.
19439 (find_taken_edge): Push sanity checking on val to workers...
19440 (find_taken_edge_cond_expr): ... here
19441 (find_taken_edge_switch_expr): ... and here, handle cases
19442 with just a default label.
19443 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19444 (group_case_labels): Likewise.
19445 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19446 group_case_labels does anything cleanup the CFG again.
19448 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19450 PR tree-optimization/81196
19451 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19452 exit condition comparing two IVs.
19454 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19456 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19457 profile to the dummy entry at the end of the list of architectures.
19458 * config/arm/arm-cpu-cdata.h: Regenerated.
19460 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19461 Michael Collison <michael.collison@arm.com>
19464 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19466 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19467 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19468 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19469 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19470 of shift when the shift amount is masked with constant equal to
19471 the size of the mode.
19472 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19475 2017-06-29 Martin Liska <mliska@suse.cz>
19477 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19479 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19482 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19485 (edge_growth_cache_entry::edge_growth_cache_entry): New
19487 (reset_edge_growth_cache): Update to use constructor.
19489 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19491 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19492 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19493 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19495 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19497 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19498 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19500 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19502 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19503 (*-linux-uclibc*): Add t-uclibc tmake_file.
19504 * config/t-musl: New.
19505 * config/t-uclibc: New.
19507 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19509 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19511 (gen_comm_data): Emit architectural setting of arch_prof.
19512 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19514 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19515 (armv8-m.base, armv8-m.main): Likewise.
19516 * arm-protos.h (arm_build_target): Add profile field.
19517 (arch_option): Likewise.
19518 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19520 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19521 arm_active_target.profile.
19523 2017-06-28 Richard Biener <rguenther@suse.de>
19525 PR middle-end/81227
19526 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19527 TYPE_OVERFLOW_WRAPS.
19528 * match.pd (negate_expr_p): Likewise.
19529 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19530 fold_build2, not fold_binary.
19532 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19534 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19535 Convert memory address to Pmode.
19536 (aarch64_print_operand): Assert MEM operands are always Pmode.
19538 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19541 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19542 Remove redundant if.
19543 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19544 * config/arm/aarch-common-protos.h
19545 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19546 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19548 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19551 * multiple_target.c (create_dispatcher_calls): Set the default
19552 clone to be static, not public.
19554 2017-06-28 Richard Biener <rguenther@suse.de>
19556 * tree-vect-loop.c (vectorizable_reduction): Move special
19557 cond reduction IV var creation ...
19558 (vect_create_epilog_for_reduction): ... here. Remove induction_index
19559 parameter. Use STMT_VINFO_VECTYPE.
19560 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
19563 2017-06-28 Martin Liska <mliska@suse.cz>
19566 * ipa-visibility.c (non_local_p): Handle visibility.
19568 2017-06-28 Martin Liska <mliska@suse.cz>
19571 * common.opt: Add IntegerRange to various options.
19572 * opt-functions.awk (integer_range_info): New function.
19573 * optc-gen.awk: Add integer_range_info to cl_options struct.
19574 * opts-common.c (decode_cmdline_option): Handle
19575 CL_ERR_INT_RANGE_ARG.
19576 (cmdline_handle_error): Likewise.
19577 * opts.c (print_filtered_help): Show valid interval in
19578 when --help is provided.
19579 * opts.h (struct cl_option): Add range_min and range_max fields.
19580 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
19582 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
19584 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
19585 (x * C EQ/NE y * C): New transformation.
19587 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
19589 * genmultilib (combination_space): Accept '+' in option names.
19591 2017-06-28 Martin Liska <mliska@suse.cz>
19594 * asan.c (instrument_derefs): Bail out inner references
19595 that are hard register variables.
19597 2017-06-28 Jakub Jelinek <jakub@redhat.com>
19600 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
19601 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
19603 2017-06-28 Richard Biener <rguenther@suse.de>
19605 * tree-vectorizer.h (vect_get_vec_defs): Remove.
19606 (vect_get_slp_defs): Adjust.
19607 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
19609 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
19611 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
19612 get_initial_defs_for_reduction instead of vect_get_vec_defs.
19613 (vectorizable_reduction): Adjust.
19614 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
19616 (vect_get_slp_defs): Likewise.
19617 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
19618 (vectorizable_bswap): Adjust.
19619 (vectorizable_call): Likewise.
19620 (vectorizable_conversion): Likewise.
19621 (vectorizable_assignment): Likewise.
19622 (vectorizable_shift): Likewise.
19623 (vectorizable_operation): Likewise.
19624 (vectorizable_store): Likewise.
19625 (vectorizable_condition): Likewise.
19626 (vectorizable_comparison): Likewise.
19628 2017-06-28 Michael Collison <michael.collison@arm.com>
19631 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
19633 (arm_gen_movmemqi): Removed unused variable 'i'.
19634 Convert 'for' loop into 'while' loop.
19635 (arm_expand_prologue): Remove last unnecessary set of insn.
19636 (thumb_pop): Remove unused variable 'pushed_words'.
19637 (thumb_exit): Remove last unnecessary set of regs_to_pop.
19639 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19641 * config/s390/predicates.md: Use s390_rel_address_ok_p.
19642 * config/s390/s390-protos.h: Add prototype of
19643 s390_rel_address_ok_p.
19644 * config/s390/s390.c (s390_got_symbol): New function.
19645 (s390_rel_address_ok_p): New function.
19646 (legitimize_pic_address): Use s390_rel_address_ok_p.
19647 (s390_load_got): Use s390_got_symbol.
19648 (s390_option_override): Issue error if
19649 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
19650 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
19652 * config/s390/s390.opt: New option mpic-data-is-text-relative.
19654 2017-06-27 Andrew Pinski <apinski@cavium.com>
19656 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
19657 (X * copysign (1.0, X)): New pattern.
19658 (X * copysign (1.0, -X)): New pattern.
19659 (copysign (-1.0, CST)): New pattern.
19661 2017-06-27 Joseph Myers <joseph@codesourcery.com>
19663 * genmultilib (combination_space): Remove variable.
19664 Validate reuse rules against regular expression for any sequence
19665 of multilib options in any order.
19667 2017-06-27 Michael Collison <michael.collison@arm.com>
19669 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
19670 call aarch64_split_simd_combine.
19671 * (aarch64_combine_internal<mode>): Delete pattern.
19672 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
19673 Allow register and subreg operands.
19675 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19677 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
19678 specific need, just fallback on defaults.
19679 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
19681 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19682 Olivier Hainque <hainque@adacore.com>
19684 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
19686 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
19687 targets. Pick a default if no particular attempt applied.
19688 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
19691 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19693 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
19694 (x86_64-wrs-vxworks7): Likewise.
19696 2017-06-27 Marek Polacek <polacek@redhat.com>
19699 * ubsan.c (instrument_null): Check get_base_address's result for null.
19701 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19703 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
19705 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19707 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
19708 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
19709 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
19710 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
19711 New function types.
19712 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
19713 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
19714 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
19715 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
19716 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
19717 BUILT_IN_FEUPDATEENV): New builtins.
19718 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
19719 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
19720 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
19721 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
19723 (builtin_structptr_types): Adjust size.
19724 * tree.c (builtin_structptr_types): Add four entries.
19726 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19727 Olivier Hainque <hainque@adacore.com>
19729 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
19730 (TLS_SYM): New local macro, forcing reference to __tls__ on
19731 link command lines for VxWorks 7 RTPs, triggering initialization
19733 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
19734 OS features TLS support, true for RTPs on VxWorks 7.
19735 * config/vxworks.c (vxworks_override_options): Setup emutls
19738 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19740 * predict.c (test_prediction_value_range): Use -1U instead of -1
19741 to avoid narrowing conversion warning.
19742 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19743 to avoid narrowing conversion warning.
19744 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19746 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19748 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19750 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19751 64bit configurations.
19752 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19753 (SIZE_TYPE): Likewise.
19754 * config/vxworks.c (vxworks_emutls_var_fields): Use
19755 long_unsigned_type_node instead of unsigned_type_node as the offset
19756 field type, which is "pointer" mode in emutls.c.
19758 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19761 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19763 PR middle-end/81207
19764 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19765 gimple_vuse copying separately from gimple_vdef copying.
19767 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19769 * value-prof.c (free_hist): Remove call to memset and the enclosing if
19772 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19773 Olivier Hainque <hainque@adacore.com>
19775 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
19776 for all vxworks7 targets.
19777 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
19778 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
19779 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
19780 variations for VX6/VX7 and 32/64bits later on in ...
19781 (VXWORKS_LIB_SPEC): Leverage new macros.
19782 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
19783 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
19785 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19787 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
19788 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
19790 2017-06-26 Carl Love <cel@us.ibm.com>
19792 * config/rs6000/rs6000-c.c: Add support for built-in functions
19793 vector bool char vec_reve (vector bool char);
19794 vector signed char vec_reve (vector signed char);
19795 vector unsigned char vec_reve (vector unsigned char);
19796 vector bool int vec_reve (vector bool int);
19797 vector signed int vec_reve (vector signed int);
19798 vector unsigned int vec_reve (vector unsigned int);
19799 vector bool long long vec_reve (vector bool long long);
19800 vector signed long long vec_reve (vector signed long long);
19801 vector unsigned long long vec_reve (vector unsigned long long);
19802 vector bool short vec_reve (vector bool short);
19803 vector signed short vec_reve (vector signed short);
19804 vector double vec_reve (vector double);
19805 vector float vec_reve (vector float);
19806 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
19807 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
19808 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
19809 (altivec_vreve): New pattern.
19810 * config/rs6000/altivec.h (vec_reve): New define.
19811 * doc/extend.texi (vec_rev): Update the built-in documentation file
19812 for the new built-in functions.
19814 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19816 PR tree-optimization/71815
19817 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19819 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19821 (slsr_process_phi): Likewise.
19822 (replace_uncond_cands_and_profitable_phis): Don't replace a
19823 multiply candidate with a stride of 1 (copy or cast).
19824 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19826 (lowest_cost_path): Likewise.
19827 (total_savings): Likewise.
19829 2017-06-26 Richard Biener <rguenther@suse.de>
19832 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19833 Use def_builtin_pure for all gather builtins.
19835 2017-06-26 Richard Biener <rguenther@suse.de>
19837 PR tree-optimization/81203
19838 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19839 non-dominating BBs.
19841 2017-06-26 Marek Polacek <polacek@redhat.com>
19844 * doc/invoke.texi: Document -Wmultistatement-macros.
19846 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19848 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19849 arm_neon_ok_no_float_abi effective target.
19851 2017-06-26 Richard Biener <rguenther@suse.de>
19853 PR tree-optimization/80928
19854 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19855 (copy_bbs): Set BB_DUPLICATED flag early.
19856 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19858 (execute_on_shrinking_pred): Likewise.
19859 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19860 BB_DUPLICATED blocks.
19861 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19862 iterate over all PHIs considering removal of *gsi.
19864 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19866 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19869 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19871 * config/rs6000/rs6000-string.c: (expand_block_clear,
19872 do_load_for_compare, select_block_compare_mode,
19873 compute_current_alignment, expand_block_compare,
19874 expand_strncmp_align_check, expand_strn_compare,
19875 expand_block_move, rs6000_output_load_multiple)
19876 Move functions related to string/block move/compare
19877 to a separate file.
19878 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19879 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19880 for this function which is now used in two files.
19881 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19882 * config.gcc: Add rs6000-string.o to extra_objs for
19883 targets powerpc*-*-* and rs6000*-*-*.
19885 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19888 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19889 32-bit, since indexed is not valid for DImode.
19890 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19891 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19892 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19893 (define_peephole2 for Altivec d-form store): Likewise.
19896 * multiple_target.c (create_dispatcher_calls): Only create the
19897 dispatcher call if the function is the default clone of a
19898 versioned function.
19900 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19902 PR middle-end/80902
19903 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19904 a call, force the call to not be a tail call.
19906 2017-06-23 Jeff Law <law@redhat.com>
19908 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19911 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19913 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19914 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19915 handling for early expansion of vector shifts (sl,sr,sra,rl).
19916 (builtin_function_type): Add vector shift right instructions
19917 to the unsigned argument list.
19919 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19921 rtl-optimizatoin/79286
19922 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19923 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19924 trap. PIC register plus a const unspec without offset can never trap.
19926 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19928 * tree.h (builtin_structptr_type): New type.
19929 (builtin_structptr_types): Declare new array.
19930 * tree.c (builtin_structptr_types): New array.
19931 (free_lang_data, build_common_tree_nodes): Use it.
19933 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19936 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19939 2017-06-22 Matt Turner <mattst88@gmail.com>
19941 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19942 Lake models to skylake case. Assume skylake for unknown
19943 models with clflushopt.
19945 2017-06-22 Jeff Law <law@redhat.com>
19947 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19948 frame sizes that do not satisfy aarch64_uimm12_shift.
19950 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19952 * profile-count.h (apply_probability,
19953 apply_scale, probability_in): Fix checks for zero.
19955 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19957 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19958 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19960 2016-06-22 Richard Biener <rguenther@suse.de>
19962 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19963 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19964 REDUC_MAX_EXPR support.
19965 (vectorizable_reduction): Likewise.
19966 (vect_create_epilog_for_reduction): Likewise.
19968 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19970 * match.pd (A / (1 << B) -> A >> B): New.
19971 * generic-match-head.c: Include optabs-tree.h.
19972 * gimple-match-head.c: Likewise.
19973 * optabs-tree.h (target_supports_op_p): New.
19974 * optabs-tree.c (target_supports_op_p): New.
19976 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19978 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19979 $gcc_cv_ld --help output.
19980 (gcc_cv_ld_demangle): Likewise.
19981 (gcc_cv_ld_eh_frame_hdr): Likewise.
19982 (gcc_cv_ld_pie): Likewise.
19983 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19984 (gcc_cv_ld_buildid): Likewise.
19985 (gcc_cv_ld_sysroot): Likewise.
19986 (ld_bndplt_support): Likewise.
19987 (ld_pushpopstate_support): Likewise.
19988 * configure: Regenerate.
19989 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19991 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19994 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19995 operands indexes to avoid gap between operands and match_dups.
19997 2017-06-21 Andrew Pinski <apinski@cavium.com>
19999 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20000 Increment Arith_shift and Arith_shift_reg by 1.
20001 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20003 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20004 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20005 (aarch64_strip_extend): Add new argument and test for it.
20006 (aarch64_cheap_mult_shift_p): New function.
20007 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20008 add a cost if it is true.
20009 Update calls to aarch64_strip_extend.
20010 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20012 2017-06-21 Andrew Pinski <apinski@cavium.com>
20014 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20016 (thunderxt88): Likewise.
20017 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20018 (thunderx_prefetch_tune): New variable.
20019 (thunderx2t99_prefetch_tune): Update for the correct values.
20020 (thunderxt88_tunings): New variable.
20021 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20022 generic_prefetch_tune.
20023 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20025 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20027 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20028 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20029 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20030 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20031 (aarch64_atomic_cas<mode>, GPI): Likewise.
20033 2017-06-21 Martin Liska <mliska@suse.cz>
20035 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20036 statements on cold and hot labels.
20037 * predict.c (tree_estimate_probability_bb): Remove the
20038 prediction from this place.
20040 2017-06-21 Martin Liska <mliska@suse.cz>
20042 PR tree-optimization/79489
20043 * gimplify.c (maybe_add_early_return_predict_stmt): New
20045 (gimplify_return_expr): Call the function.
20046 * predict.c (tree_estimate_probability_bb): Remove handling
20048 * predict.def: Update comment about early return predictor.
20049 * gimple-predict.h (is_gimple_predict): New function.
20050 * predict.def: Change default value of early return to 66.
20051 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20053 * passes.def: Put pass_strip_predict_hints to the beginning of
20056 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20058 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20059 FUNCTION_DECL declarations.
20060 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20062 (dwaf2out_decl): Likewise.
20063 * godump.c (go_early_global_decl): Skip call to the real debug hook
20064 for FUNCTION_DECL declarations.
20065 * passes.c (rest_of_decl_compilation): Skip call to the
20066 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20067 -fdump-go-spec is passed.
20069 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20071 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20072 Reorder for compactness.
20073 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20074 (def_builtin_pure, def_builtin_pure2): New functions.
20075 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20077 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20079 * match.pd (nop_convert): New predicate.
20080 ((A +- CST1) +- CST2): Allow some NOP conversions.
20082 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20085 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20086 with ctors/dtors if GOVD_SHARED is set.
20088 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20090 * config/aarch64/aarch64.md (movti_aarch64):
20091 Emit mov rather than orr.
20092 (movtf_aarch64): Likewise.
20093 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20094 Emit mov rather than orr.
20096 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20098 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20099 Swap alternatives, make integer dup more expensive.
20101 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20103 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20104 Return true for non-tls symbols.
20106 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20108 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20109 (cortex-a75): Likewise.
20110 (cortex-a75.cortex-a55): Likewise.
20111 * config/aarch64/aarch64-tune.md: Regenerate.
20112 * doc/invoke.texi (-mtune): Document new values for -mtune.
20114 2017-06-21 Tom de Vries <tom@codesourcery.com>
20116 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20117 stack_size feature.
20118 (Effective-Target Keywords, Other attributes): Suggest using
20119 dg-add-options stack_size feature to get stack limit in stack_size
20120 effective target documentation.
20122 2017-06-21 Julian Brown <julian@codesourcery.com>
20123 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20125 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20126 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20127 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20129 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20130 attribute type list for neon_multiply.
20131 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20132 attribute type list for neon_multiply.
20133 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20134 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20135 attribute type list for neon_multiply.
20136 * config/arm/types.md (crypto_pmull): Add.
20137 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20138 attribute type list.
20140 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20142 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20145 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20147 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20148 to make sure not to dereference a NULL cost_classes_ptr pointer.
20150 2017-06-20 Carl Love <cel@us.ibm.com>
20152 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20153 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20154 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20155 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20156 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20157 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20158 VMULOSW): New enum "unspec" values.
20159 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20160 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20161 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20162 altivec_vmulosw): New patterns.
20163 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20164 VMULOSW): Add definitions.
20166 2017-06-20 Julia Koval <julia.koval@intel.com>
20168 * config/i386/i386.c: Fix rounding expand for new pattern.
20169 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20171 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20173 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20174 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20176 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20178 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20181 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20183 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20184 architecture, then by implementer ID.
20185 * config/aarch64/aarch64-tune.md: Regenerate.
20187 2017-06-20 Richard Biener <rguenther@suse.de>
20189 PR middle-end/81097
20190 * fold-const.c (split_tree): Fold to type before negating.
20192 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20194 * diagnostic-show-locus.c
20195 (selftest::test_fixit_deletion_affecting_newline): New function.
20196 (selftest::diagnostic_show_locus_c_tests): Call it.
20198 2017-06-20 Andreas Schwab <schwab@suse.de>
20201 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20204 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20206 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20207 __ARM_FEATURE_COPROC according to support.
20209 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20211 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20212 Rewritten to avoid overflow for > 32-bit pointers.
20215 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20216 by removing enum keyword.
20217 (ubsan_type_descriptor): Likewise. Formatting fix.
20220 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20221 splitter): Require TARGET_SSE2 in the condition.
20223 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20226 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20227 for doing vector set of SFmode on ISA 3.0.
20228 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20229 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20231 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20232 SFmode value into a V4SF variable that was extracted from another
20233 V4SF variable without converting the element to double precision
20234 and back to single precision vector format.
20235 (vsx_insert_extract_v4sf_p9_2): Likewise.
20237 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20239 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20240 in UWHI to avoid undefined overflow.
20243 * ubsan.h (enum ubsan_encode_value_phase): New.
20244 (ubsan_encode_value): Change second argument to
20245 enum ubsan_encode_value_phase with default value of
20246 UBSAN_ENCODE_VALUE_GENERIC.
20247 * ubsan.c (ubsan_encode_value): Change second argument to
20248 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20249 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20250 create_tmp_var_raw instead of create_tmp_var and use a
20252 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20253 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20254 ubsan_encode_value callers.
20257 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20258 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20259 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20261 2017-06-19 Richard Biener <rguenther@suse.de>
20263 PR middle-end/81118
20264 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20265 estimates if we changed anything.
20267 2017-06-19 Richard Biener <rguenther@suse.de>
20269 PR tree-optimization/80887
20270 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20271 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20272 simplified lookups, then reset mprts_hook.
20273 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20275 (try_to_simplify): Likewise.
20277 2017-06-19 Martin Liska <mliska@suse.cz>
20280 * gimplify.c (gimplify_switch_expr):
20281 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20283 2017-06-19 Martin Liska <mliska@suse.cz>
20285 * doc/install.texi: Document that PGO runs in 4 stages.
20287 2017-06-19 Martin Liska <mliska@suse.cz>
20290 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20291 to dispatcher function name.
20292 * multiple_target.c (replace_function_decl): New function.
20293 (create_dispatcher_calls): Redirect both edges and references.
20295 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20297 * profile-count.c (profile_count::dump): Dump quality.
20298 (profile_count::differs_from_p): Update for unsigned val.
20299 * profile-count.h (profile_count_quality): New enum.
20300 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20302 2017-06-19 Richard Biener <rguenther@suse.de>
20304 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20305 struct function as arg.
20306 (estimate_numbers_of_iterations): Export overload with loop arg.
20307 (free_numbers_of_iterations_estimates_loop): Use an overload of
20308 free_numbers_of_iterations_estimates instead.
20309 * tree-cfg.c (remove_bb): Adjust.
20310 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20311 * tree-parloops.c (gen_parallel_loop): Likewise.
20312 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20314 (tree_unroll_loops_completely): Likewise.
20315 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20316 Use an overload instead and export.
20317 (estimated_loop_iterations): Adjust.
20318 (max_loop_iterations): Likewise.
20319 (likely_max_loop_iterations): Likewise.
20320 (estimate_numbers_of_iterations): Take struct function as arg
20322 (loop_exits_before_overflow): Adjust.
20323 (free_numbers_of_iterations_estimates_loop): Use an overload.
20324 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20325 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20327 2017-06-19 Richard Biener <rguenther@suse.de>
20330 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20331 RANGE_EXPR conservatively.
20333 2017-06-16 Carl Love <cel@us.ibm.com>
20335 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20336 definitions for vec_float, vec_float2, vec_floato,
20337 vec_floate built-ins.
20338 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20339 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20341 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20342 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20343 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20344 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20345 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20346 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20347 vec_floato): Add builtin defines.
20348 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20349 Update the built-in documentation file for the new built-in
20352 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20354 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20355 (mthumb): Mark as the negative of -marm.
20357 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20359 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20361 (ARM Options, -mtune): Document that this accepts the same
20362 extension options as -mcpu.
20363 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20365 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20367 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20368 permitted extensions.
20370 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20372 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20373 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20374 (armv8-m.main): Add option +nodsp.
20375 * config/arm/arm-cpu-cdata.h: Regenerated.
20377 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20379 * config/arm/t-fuchsia: New file.
20380 * config.gcc (arm*-*-fuchsia*): Use it.
20382 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20384 * config/arm/t-symbian: Rewrite for new option infrastructure.
20386 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20388 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20389 (MULTILIB_REQUIRED): Likewise.
20391 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20393 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20394 (MULTILIB_RESUE): Likewise.
20395 (MULTILIB_MATCHES): Likewise.
20396 (MULTLIB_REQUIRED): Likewise.
20398 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20400 * config/arm/t-rtems: Rewrite for new option framework.
20402 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20404 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20405 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20406 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20407 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20408 * config/arm/t-multilib: ... here.
20409 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20410 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20411 armv7-a and armv8*-a when A-profile libraries have not been built.
20412 * config/arm/t-rmprofile: Rewrite.
20414 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20416 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20417 with a backslash. Remove the backslash after substituting unescaped
20419 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20421 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20423 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20424 the driver to pass through the default setting of -mfloat-abi.
20425 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20427 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20428 (all_feat_combs): New rule.
20429 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20431 * config/arm/t-aprofile: Rewrite.
20433 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20435 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20436 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20437 fpu is not specified by the user/command-line.
20438 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20439 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20440 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20441 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20442 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20443 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20445 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20447 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20448 * config/arm/t-arm-elf: Rewritten.
20450 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20452 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20453 have some floating-point instructions.
20454 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20455 (TARGET_MAYBE_HARD_FLOAT): New macro.
20456 * config/arm/arm-builtins.c (arm_init_builtins): Use
20457 TARGET_MAYBE_HARD_FLOAT.
20458 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20460 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20462 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20463 (configargs.h): Include it.
20464 (arm_print_hint_for_fpu_option): New function.
20465 (arm_parse_fpu_option): New function.
20466 (candidate_extension): New class.
20467 (arm_canon_for_multilib): New function.
20468 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20469 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20470 (ARCH_CANONICAL_SPECS): New macro.
20471 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20473 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20475 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20476 are set after handling multilib fragments. Set target_cpu_default2
20479 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20481 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20483 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20484 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20486 (check_arch): Likewise.
20487 * config/arm/arm.c (arm_configure_build_target): Handle
20488 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20489 options in the default.
20491 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20493 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20494 when an option is an alias of another.
20495 * config/arm/parsecpu.awk (optalias): New parser token.
20496 (gen_comm_data): Mark non-alias options as such. Emit entries
20497 for extension aliases.
20498 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20499 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20500 (armv6kz, armv6zk, armv6t2): Likewise.
20501 (armv7): Make vfpv3-d16 an alias.
20502 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20504 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20505 Sort in canonical order.
20506 (armv8-a): Sort in canonical order.
20507 (armv8.1-a, armv8.2-a): Likewise.
20508 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20510 (cortex-a9): Sort in canonical order.
20511 * config/arm/arm.c (selftests.h): Include it.
20512 (arm_test_cpu_arch_data): New function.
20513 (arm_run_self_tests): New function.
20514 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20515 (targetm): Move declaration to the end of the file.
20516 * arm-cpu-cdata.h: Regenerated.
20518 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20520 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20521 call to target_mode_check describing the type of option passed.
20522 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20523 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20524 arm_parse_cpu_option_name to match parameters against list of
20526 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20527 arm_arch_core_flags data structure.
20528 * config/arm/arm-cpu_cdata.h: Regenerated.
20530 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20532 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20534 (arm_print_hint_for_cpu_option): Likewise.
20535 (arm_print_hint_for_arch_option): Likewise.
20536 (arm_parse_cpu_option_name): Likewise.
20537 (arm_parse_arch_option_name): Likewise.
20538 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20539 of entries in the all_fpus list.
20540 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20541 (arm_parse_cpu_option_name): Declare.
20542 (arm_parse_arch_option_name): Declare.
20543 (arm_parse_option_features): Declare.
20544 (arm_intialize_isa): Declare.
20545 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20547 (gen_comm_data): ... here. Make definitions non-static.
20548 * config/arm/arm-cpu-data.h: Regenerated.
20549 * config/arm/arm-cpu-cdata.h: Regenerated.
20551 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20553 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20554 (cpu_arch_extension): New structure.
20555 (cpu_arch_option, arch_option, cpu_option): New structures.
20556 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
20557 architecture types.
20558 (gen_data): Generate new format data tables.
20559 * config/arm/arm.c (cpu_tune): New structure.
20560 (cpu_option, processors): Delete.
20561 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
20562 (arm_print_hint_for_cpu_option): ... this and ...
20563 (arm_print_hint_for_arch_option): ... this.
20564 (arm_parse_arch_cpu_name): Delete. Replace with ...
20565 (arm_parse_cpu_option_name): ... this and ...
20566 (arm_parse_arch_option_name): ... this.
20567 (arm_unrecognized_feature): Change type of target parameter to
20569 (arm_parse_arch_cpu_features): Delete. Replace with ...
20570 (arm_parse_option_features): ... this.
20571 (arm_configure_build_target): Rework to use new configuration data
20573 (arm_print_tune_info): Rework for new configuration data tables.
20574 * config/arm/arm-cpu-data.h: Regenerated.
20575 * config/arm/arm-cpu.h: Regenerated.
20577 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20579 * Makefile.in (OBJS): Move sbitmap.o from here ...
20580 (OBJS-libcommon): ... to here.
20582 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20584 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
20585 (ISA_ALL_CRYPTO): New macro.
20586 (ISA_ALL_SIMD): New macro
20587 (ISA_ALL_FP): New macro.
20588 * config/arm/arm.c (fpu_bitlist): Update initializer.
20589 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
20591 (arm9e): Add fpu. Add option for nofp
20592 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
20593 (arm926ej-s, arm1026ej-s): Likewise.
20594 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
20595 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
20596 neon-fp16, neon-vfpv4, nofp and nosimd.
20597 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
20598 (cortex-a8): Add fpu. Add option for nofp.
20599 (cortex-a9): Add fpu. Add options for nosimd and nofp.
20600 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
20601 (cortex-r4f): Add fpu.
20602 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
20603 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
20605 (cortex-r8): Likewise.
20606 (cortex-m4): Add fpu. Add option for nofp.
20607 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
20608 (cortex-a17.cortex-a7): Likewise.
20609 (cortex-a32): Add fpu. Add options for crypto and nofp.
20610 (cortex-a35, cortex-a53): Likewise.
20611 (cortex-a57): Add fpu. Add option for crypto.
20612 (cortex-a72, cortex-a73): Likewise.
20613 (exynos-m1): Likewise.
20614 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
20615 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
20616 (cortex-m33): Add fpu. Add option for nofp.
20617 * config/arm/arm-cpu-cdata.h: Regenerated
20618 * config/arm/arm-cpu-data.h: Regenerated.
20620 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20622 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
20623 (armv5te, armv5tej): Likewise.
20624 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
20625 (armv7): Add options fp and vfpv3-d16.
20626 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
20627 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
20629 (armv7ve): Likewise.
20630 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
20631 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
20632 (armv8-a): Add nocrypto option.
20633 (armv8.1-a, armv8.2-a): Likewise.
20634 (armv8-m.main): add options fp, fp.dp and nofp.
20636 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20638 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
20640 (armv8-a+crc): Delete.
20641 (armv8.1-a): Add options simd, crypto and nofp.
20642 (armv8.2-a): Add options fp16, simd, crypto and nofp.
20643 (armv8.2-a+fp16): Delete.
20644 (armv8-m.main): Add option dsp.
20645 (armv8-m.main+dsp): Delete.
20646 (cortex-a8): Add fpu. Add nofp option.
20647 (cortex-a9): Add fpu. Add nofp and nosimd options.
20648 * config/arm/parsecpu.awk (gen_data): Generate option tables and
20649 link to main cpu and architecture data structures.
20650 (gen_comm_data): Only put isa attributes from the main architecture
20652 (option): New statement for architecture and CPU entries.
20653 * arm.c (struct cpu_option): New structure.
20654 (struct processors): Add entry for options.
20655 (arm_unrecognized_feature): New function.
20656 (arm_parse_arch_cpu_name): Ignore any characters after the first
20658 (arm_parse_arch_cpu_feature): New function.
20659 (arm_configure_build_target): Separate out any CPU and architecture
20660 features and parse separately. Don't error out if -mfpu=auto is
20661 used with only an architecture string.
20662 (arm_print_asm_arch_directives): New function.
20663 (arm_file_start): Call it.
20664 * config/arm/arm-cpu-cdata.h: Regenerated.
20665 * config/arm/arm-cpu-data.h: Likewise.
20666 * config/arm/arm-tables.opt: Likewise.
20668 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20670 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
20671 assembler when it is not -mfpu=auto.
20673 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20675 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
20676 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
20677 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
20678 (ASM_CPU_SPEC): Rewrite.
20679 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
20680 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
20681 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
20682 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
20683 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
20684 copied string is NUL-terminated. Also strip any characters prefixed
20686 (arm_rewrite_selected_arch): New function.
20687 (arm_rewrite_march): New function.
20689 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20691 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
20692 (x_arm_cpu_string, x_arm_tune_string): Likewise.
20693 (march, mcpu, mtune): Convert to string-based options.
20694 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
20695 (arm_parse_arch_cpu_name): New function.
20696 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
20697 identify selected architecture or CPU.
20698 (arm_option_save): New function.
20699 (TARGET_OPTION_SAVE): Redefine.
20700 (arm_option_restore): Restore string options.
20701 (arm_option_print): Print string options.
20703 2017-06-16 Martin Sebor <msebor@redhat.com>
20705 PR tree-optimization/80933
20706 PR tree-optimization/80934
20707 * builtins.c (fold_builtin_3): Do not handle bcmp here.
20708 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
20709 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
20710 (gimple_fold_builtin): Call them.
20712 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20714 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
20715 as unlikely; update profile.
20717 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20719 * predict.c (force_edge_cold): Handle declaring edges impossible
20722 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20724 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
20726 (try_unroll_loop_completely): Fix reporting.
20728 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20730 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
20732 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
20735 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
20736 if given a non-constant argument for an intrinsic which requires a
20739 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20741 * profile.c (compare_freqs): New function.
20742 (branch_prob): Sort edge list.
20743 (find_spanning_tree): Assume that the list is priority sorted.
20745 2017-06-16 Richard Biener <rguenther@suse.de>
20747 PR tree-optimization/81090
20748 * passes.def (pass_record_bounds): Remove.
20749 * tree-pass.h (make_pass_record_bounds): Likewise.
20750 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20751 make_pass_record_bounds): Likewise.
20752 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20753 not free niter estimates at the beginning but at the end.
20754 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20756 2017-06-16 Richard Biener <rguenther@suse.de>
20758 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20759 initializer to workaround ICE in host GCC 4.8.
20761 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20763 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20765 (clone_inlined_nodes): Update.
20767 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20769 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
20770 prefetch settings, and enable prefetching by default at -O3.
20772 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20774 * config/aarch64/aarch64.c (aarch64_override_options_internal):
20775 Set flag_prefetch_loop_arrays according to tuning data.
20777 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20779 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
20780 New tune structure.
20781 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
20782 [Unrelated to main purpose of the patch] Place the pointer field last
20783 to enable type checking errors when tune structure are wrongly merged.
20784 * config/aarch64/aarch64.c (generic_prefetch_tune,)
20785 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
20786 (thunderx2t99_prefetch_tune): New tune constants.
20787 (tune_params *_tunings): Update all tunings (no functional change).
20788 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
20789 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
20790 from tunings structures.
20792 2017-06-16 Jakub Jelinek <jakub@redhat.com>
20795 * ubsan.c (instrument_null): Add T argument, use it instead
20796 of computing it based on IS_LHS.
20797 (instrument_object_size): Likewise.
20798 (pass_ubsan::execute): Adjust instrument_null and
20799 instrument_object_size callers to pass gimple_get_lhs or
20800 gimple_assign_rhs1 result to it. Use instrument_null instead of
20801 calling get_base_address and instrument_mem_ref. Handle
20802 aggregate call arguments for object-size sanitization.
20804 2017-06-16 Yury Gribov <tetra2005@gmail.com>
20806 PR tree-optimization/81089
20807 * tree-vrp.c (is_masked_range_test): Validate operands of
20810 2017-06-15 Martin Sebor <msebor@redhat.com>
20813 * dumpfile.c (dump_register): Avoid calling memset to initialize
20814 a class with a default ctor.
20815 * gcc.c (struct compiler): Remove const qualification.
20816 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
20817 * hash-table.h: Ditto.
20818 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20820 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20821 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20823 * params.h (struct param_info): Make struct members non-const.
20824 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20825 with default initialization.
20826 * vec.h (vec_copy_construct, vec_default_construct): New helper
20828 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20829 with vec_copy_construct.
20830 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20831 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20832 * doc/invoke.texi (-Wclass-memaccess): Document.
20834 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20836 * emit-rtl.h (is_leaf): Update comment about local
20837 register allocator.
20839 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20842 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20843 for a variable to have a section before checking if the section has a
20845 Set section to.persistent if persistent attribute is set.
20846 Warn if .persistent attribute is used on an automatic variable.
20848 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20850 PR rtl-optimization/80474
20851 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20853 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20855 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20858 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20860 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20861 (casesi): Emit bounds checking as RTL.
20862 (casesi_internal_mips16_<mode>): Remove bounds checking.
20864 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20866 * config/xtensa/xtensa.c (xtensa_option_override): Append
20867 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20868 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20869 xtensa_doloop_hooks): Define unconditionally.
20870 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20872 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20873 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20874 for it in xtensa_option_override.
20875 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20876 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20878 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20880 * doc/cppopts.texi: Document '-' special value to -MF.
20882 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20884 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20885 (cortex_a53_fconst): Likewise.
20886 (cortex_a53_fpmul): Likewise.
20887 (cortex_a53_f_load_64): Likewise.
20888 (cortex_a53_f_load_many): Likewise.
20889 (cortex_a53_advsimd_alu): Likewise.
20890 (cortex_a53_advsimd_alu_q): Likewise.
20891 (cortex_a53_advsimd_mul): Likewise.
20892 (cortex_a53_advsimd_mul_q): Likewise.
20893 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20894 Add missing fmul, r2f_cvt and fconst cases.
20896 2017-06-14 Richard Biener <rguenther@suse.de>
20898 PR middle-end/81088
20899 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20901 (fold_binary_loc): When associating do not treat pre-existing
20902 TREE_OVERFLOW on literal constants as a reason to allow
20903 TREE_OVERFLOW on associated literal constants.
20905 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20907 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20908 (MASK_FEATURES): New macro.
20909 * config/sparc/sparc.c (sparc_option_override): Remove the special
20910 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20912 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20914 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20915 a division of 0 if non-call exceptions are enabled.
20917 2017-06-14 Andrew Pinski <apinski@cavium.com>
20918 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20921 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20922 Improve vector initialization code gen for only variable case.
20924 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20926 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20928 2017-06-14 Richard Biener <rguenther@suse.de>
20930 PR tree-optimization/81083
20931 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20934 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20936 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20937 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20938 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20939 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20940 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20941 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20943 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20945 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20946 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20948 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20950 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20952 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20954 * config/rs6000/t-rtems: Don't handle SPE.
20956 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20958 * config/rs6000/t-linux: Don't handle SPE.
20960 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20962 * config/rs6000/eabispe.h: Delete file.
20964 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20966 * config/rs6000/t-spe: Delete file.
20968 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20970 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20971 (rs6000_legitimate_offset_address_p): Return false for anything in
20972 V2SImode or V2SFmode.
20974 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20976 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20977 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20978 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20980 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20981 (rs6000_legitimate_offset_address_p): Ditto.
20982 (rs6000_emit_move): Ditto.
20983 (rs6000_init_builtins): Remove V4HI_type_node.
20985 2017-06-13 Martin Liska <mliska@suse.cz>
20988 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20989 (gate_asan): Likewise.
20990 * asan.h (asan_no_sanitize_address_p): Remove the function.
20991 (sanitize_flags_p): New function.
20992 * builtins.def: Fix coding style.
20993 * common.opt: Use renamed enum value.
20994 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20995 * doc/extend.texi: Document no_sanitize attribute.
20996 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20997 to SANITIZE_UNDEFINED_NONDEFAULT.
20998 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20999 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21000 Use sanitize_flags_p.
21001 * gimplify.c (gimplify_function_tree): Likewise.
21002 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21003 * opts.c (parse_no_sanitize_attribute): New function.
21004 (common_handle_option): Use renamed enum value.
21005 * opts.h (parse_no_sanitize_attribute): Declare.
21006 * tree.c (sanitize_flags_p): New function.
21007 * tree.h: Declared here.
21008 * tsan.c: Use sanitize_flags_p.
21009 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21010 (instrument_mem_ref): Likewise.
21011 (instrument_bool_enum_load): Likewise.
21012 (do_ubsan_in_current_function): Remove the function.
21013 (pass_ubsan::execute): Use sanitize_flags_p.
21014 * ubsan.h: Remove do_ubsan_in_current_function
21015 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21016 (dump_function_to_file): Use it here.
21018 2017-06-13 Martin Jambor <mjambor@suse.cz>
21020 PR tree-optimization/80803
21021 PR tree-optimization/81063
21022 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21023 (propagate_subaccesses_across_link): Enqueue subtree whenever
21024 necessary instead of relying on the caller.
21026 2017-06-13 Martin Jambor <mjambor@suse.cz>
21028 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21029 that have a first_link.
21030 (sort_and_splice_var_accesses): Do not check first_link before
21032 (subtree_mark_written_and_enqueue): Likewise.
21033 (propagate_all_subaccesses): Likewise and do not stop at first
21034 parent with a first_link.
21036 2017-06-13 Martin Jambor <mjambor@suse.cz>
21038 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21041 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21043 * match.pd: New pattern.
21045 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21047 * tree-vrp.c (is_masked_range_test): New function.
21048 (register_edge_assert_for): Determine ranges for
21051 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21053 PR tree-optimization/67328
21054 * fold-const.c (maskable_range_p): New function.
21055 (build_range_check): Generate bittests if possible.
21057 2017-06-13 Martin Liska <mliska@suse.cz>
21059 * gimple-pretty-print.c (dump_probability): Add new argument.
21060 (dump_edge_probability): Dump both probability and count.
21061 (dump_gimple_label): Likewise.
21062 (dump_gimple_bb_header): Likewise.
21064 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21067 * config/avr/avr-devices.c: Fix indentation.
21068 * config/avr/gen-avr-mmcu-specs.c: Dito.
21070 2017-06-13 Richard Biener <rguenther@suse.de>
21072 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21073 instead get vector type from stmt_info.
21074 (vectorizable_reduction): Adjust. Remove dead code.
21076 2017-06-13 Richard Biener <rguenther@suse.de>
21078 PR middle-end/81065
21079 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21080 case of C * (x * C2 + C3).
21081 (fold_addr_of_array_ref_difference): Properly fold index difference.
21083 2017-06-12 David S. Miller <davem@davemloft.net>
21086 * config/sparc/sparc.md (return expander): Emit frame blockage if
21087 function uses alloca.
21089 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21091 * combine.c (make_field_assignment): Check len rather than the mode
21092 precision when calling force_to_mode.
21094 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21096 Support multilibs and devices that see flash in RAM address range.
21099 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21100 (avr_mcu_t) <flash_pm_offset>: New field.
21101 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21102 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21103 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21104 (AVR_TINY_PM_OFFSET): Remove macro.
21105 * config/avr/avr.opt (-mshort-calls): New option.
21106 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21107 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21108 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21109 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21110 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21111 instead of avr_arch->have_jmp_call.
21112 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21113 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21114 avr_arch->flash_pm_offset to define.
21115 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21116 new field flash_pm_offset. Add entry for avrxmega3.
21117 (avr_texinfo): Add entry for avrxmega3.
21118 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21119 attiny212, attiny214,
21120 attiny412, attiny414, attiny416, attiny417,
21121 attiny814, attiny816, attiny817,
21122 attiny1614, attiny1616, attiny1617,
21123 attiny3214, attiny3216, attiny3217.
21124 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21125 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21126 (avr_print_operand_address) [AVR_TINY]: Same.
21127 (avr_asm_init_sections) <readonly_data_section>: Only patch
21128 callback if avr_arch->flash_pm_offset = 0.
21129 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21130 for rodata if avr_arch->flash_pm_offset != 0.
21131 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21132 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21133 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21134 (m_options): Append opt_rcall.
21135 (m_dirnames): Append dir_rcall.
21136 * config/avr/t-multilib: Regenerate.
21138 * configure.ac [target=avr]: Check whether avrxmega3 default
21139 linker description file works as needed.
21140 * configure: Regenerate.
21141 * doc/avr-mmcu.texi: Regenerate.
21142 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21143 <__AVR_ARCH__>: Document avrxmega3 and 103.
21144 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21145 <__AVR_SHORT_CALLS__>: Document it.
21146 <__AVR_PM_BASE_ADDRESS__>: Document it.
21147 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21148 (AVR Variable Attributes) <progmem>: Document this is
21149 not needed for avrxmega3.
21150 (AVR Named Address Spaces) <__flash>: Dito.
21152 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21154 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21156 2017-06-12 Doug Rupp <rupp@adacore.com>
21158 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21159 Append vxworks-stdint.h to the tm_file list.
21160 * config/vxworks-stdint.h: New file.
21162 2017-06-12 Martin Liska <mliska@suse.cz>
21164 PR tree-optimization/81041
21165 * tree-profile.c (gimple_gen_ic_func_profiler):
21166 Create an extra BB in profile-generate
21167 (gimple_gen_time_profiler): Likewise.
21169 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21171 PR tree-optimization/81003
21172 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21173 (update_range_test): Use it instead of force_gimple_operand_gsi.
21175 2017-06-12 Richard Biener <rguenther@suse.de>
21177 PR tree-optimization/81053
21178 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21179 with backedge value not defined in loop. Simplify def stmt
21182 2017-06-11 Tom de Vries <tom@codesourcery.com>
21185 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21187 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21188 nvptx_cannot_force_const_mem.
21190 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21192 * opts.c (finish_options): Move test for flag_split_stack after
21193 it has been initialized.
21195 2017-06-11 Jason Merrill <jason@redhat.com>
21197 * tree.h (id_equal): New.
21198 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21199 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21200 instead of strcmp of IDENTIFIER_POINTER.
21202 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21204 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21205 (mark_all_inlined_calls_cdtor): Fix formating.
21206 (inline_transform): Rescale profile before inlining.
21208 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21210 * cgraph.h (cgraph_edge::clone): Update prototype.
21211 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21212 (cgraph_node::create_clone): Update.
21213 (cgraph_node::create_version_clone): Update.
21214 * tree-inline.c (copy_bb): Update.
21215 (expand_call_inline): Update.
21217 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21219 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21220 factored out from ...
21221 (rs6000_emit_prologue): ... here.
21223 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21225 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21226 factored out from ...
21227 (rs6000_emit_prologue): ... here.
21229 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21231 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21233 (handle_missing_profiles): Fix computation of tp_first_run.
21234 (counts_to_freqs): Do not touch freqs when count is 0.
21236 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21238 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21241 2017-06-10 Tom de Vries <tom@codesourcery.com>
21243 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21244 attributes): Document signal effective target.
21246 2017-06-10 Tom de Vries <tom@codesourcery.com>
21248 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21249 Document effective target stack_size.
21251 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21253 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21254 to the edit_context if they can be auto-applied.
21256 2017-06-9 Ian Lance Taylor <iant@golang.org>
21258 * opts.c (finish_options): If -fsplit-stack, disable implicit
21259 -forder-blocks-and-partition.
21260 * doc/invoke.texi (Optimize Options): Document that when using
21261 -fsplit-stack -forder-blocks-and-partition is not implicitly
21264 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21266 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21267 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21268 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21269 * builtins.def (abort, trap, unreachable): Declare cold.
21270 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21271 * tree-core.h (ECF_COLD): New.
21272 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21273 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21275 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21277 * predict.c (unlikely_executed_stmt_p): Cleanup.
21279 2017-06-09 Richard Biener <rguenther@suse.de>
21281 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21282 model if the ref is always written to.
21284 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21286 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21288 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21290 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21293 2017-06-09 Tom de Vries <tom@codesourcery.com>
21296 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21297 "target cannot support label values" when encountering LABEL_REF.
21299 2017-06-09 Martin Liska <mliska@suse.cz>
21301 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21302 (gimple_gen_ic_func_profiler): Emit direct comparison
21303 of __gcov_indirect_call_callee with NULL.
21304 (gimple_gen_time_profiler): Change probability from
21305 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21307 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21309 * profile.c (edge_gcov_counts): Turn to pointer.
21310 (compute_branch_probabilities, compute_branch_probabilities): Update.
21311 (branch_prob): Do not clear edge_gcov_count.
21312 * profile.h (edge_gcov_counts): Turn to pointer.
21313 (edge_gcov_count): Update.
21315 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21317 * gimple.h (gimple_check_failed): Mark cold.
21319 2017-06-09 Richard Biener <rguenther@suse.de>
21321 PR tree-optimization/66623
21322 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21323 refactor check_reduction into two parts, properly computing
21324 whether we have to check reduction validity for outer loop
21327 2017-06-09 Richard Biener <rguenther@suse.de>
21329 PR tree-optimization/79483
21330 * graphite-scop-detection.c (order): New global.
21331 (get_order): Compute bb to order mapping that satisfies code
21332 generation constraints.
21333 (cmp_pbbs): New helper.
21334 (build_scops): Start domwalk at entry block, sort generated
21337 2017-06-09 Richard Biener <rguenther@suse.de>
21339 PR middle-end/81007
21340 * ipa-polymorphic-call.c
21341 (ipa_polymorphic_call_context::restrict_to_inner_class):
21342 Skip FIELD_DECLs with error_mark_node type.
21343 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21346 2017-06-09 Martin Liska <mliska@suse.cz>
21348 * predict.c (struct branch_predictor): New struct.
21349 (test_prediction_value_range): New test.
21350 (predict_c_tests): New function.
21351 * selftest-run-tests.c (selftest::run_tests): Run the function.
21352 * selftest.h: Declare new tests.
21354 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21357 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21358 gen_add3_insn did not fail.
21359 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21360 r0, construct that number in a temporary reg and add that reg to r0.
21361 If asked to put the result in r0 as well, fail.
21363 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21365 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21366 for early expansion of vec_eqv.
21368 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21370 PR middle-end/81005
21371 * ubsan.c (instrument_null): Avoid pointless code temporary.
21372 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21374 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21378 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21381 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21384 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21386 * predict.c (unlikely_executed_edge_p): Move ahead.
21387 (probably_never_executed_edge_p): Use it.
21389 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21391 PR middle-end/79988
21392 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21393 gimple_call_builtin_p call.
21395 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21397 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21398 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21399 rtl_check_failed_type2, rtl_check_failed_code1,
21400 rtl_check_failed_code2, rtl_check_failed_code_mode,
21401 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21402 rtvec_check_failed_bounds, rtl_check_failed_flag,
21403 _fatal_insn_not_found, _fatal_insn): Likewise.
21404 * tree.h (tree_contains_struct_check_failed,
21405 tree_check_failed, tree_not_check_failed,
21406 tree_class_check_failed, tree_range_check_failed,
21407 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21408 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21409 tree_operand_check_failed, omp_clause_check_failed,
21410 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21413 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21415 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21416 flag_branch_probabilities.
21417 * ipa-inline.c (edge_badness): Likewise.
21418 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21419 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21420 * predict.c (maybe_hot_frequency_p): Likewise.
21421 (probably_never_executed): Likewise.
21422 * sched-ebb.c (schedule_ebbs): Likewise.
21423 * sched-rgn.c (find_single_block_region): Likewise.
21424 * tracer.c (tail_duplicate): Likewise.
21426 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21428 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21429 longer requires x_flag_profile_use.
21431 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21433 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21434 instead of flag_reorder_blocks_and_partition.
21435 * dbxout.c (dbxout_function_end): Likewise.
21436 * dwarf2out.c (gen_subprogram_die): Likewise.
21437 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21438 * hw-doloop.c (reorg_loops): Likewise.
21439 * varasm.c (assemble_start_function,
21440 assemble_end_function): Likewise.
21441 (decide_function_section): Do not check for
21442 flag_reorder_blocks_and_partition.
21444 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21446 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21448 (chkp_get_hard_register_fake_addr_expr): Ditto.
21449 (chkp_build_addr_expr): Add check for hard reg case.
21450 (chkp_parse_array_and_component_ref): Ditto.
21451 (chkp_find_bounds_1): Ditto.
21452 (chkp_process_stmt): Don't generate bounds store for
21455 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21457 * predict.c (maybe_hot_bb_p): Do not check profile status.
21458 (maybe_hot_edge_p): Likewise.
21459 (probably_never_executed): Check for zero counts even if profile
21461 (unlikely_executed_edge_p): New function.
21462 (unlikely_executed_stmt_p): New function.
21463 (unlikely_executed_bb_p): New function.
21464 (set_even_probabilities): Use unlikely predicates.
21465 (combine_predictions_for_bb): Likewise.
21466 (predict_paths_for_bb): Likewise.
21467 (predict_paths_leading_to_edge): Likewise.
21468 (determine_unlikely_bbs): New function.
21469 (estimate_bb_frequencies): Use it.
21470 (compute_function_frequency): Use zero counts even if profile is
21472 * profile-count.h: Fix typo.
21474 2017-08-08 Julia Koval <julia.koval@intel.com>
21476 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21477 _mm512_mask_cvtsepi16_storeu_epi8,
21478 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21479 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21480 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21481 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21482 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21483 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21484 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21485 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21486 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21487 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21488 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21489 __builtin_ia32_pmovuswb256mem_mask,
21490 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21491 __builtin_ia32_pmovwb512mem_mask): New builtins.
21493 2017-08-08 Julia Koval <julia.koval@intel.com>
21495 PR target/73350,80862
21496 * config/i386/subst.md (round): Fix round pattern.
21497 * config/i386/i386.c (ix86_erase_embedded_rounding):
21498 Fix erasing rounding for the fixed pattern.
21500 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21502 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21504 2017-06-08 Martin Liska <mliska@suse.cz>
21506 PR gcov-profile/80911
21507 * gcov.c (block_info::block_info): New constructor.
21509 2017-06-07 Carl Love <cel@us.ibm.com>
21511 * config/rs6000/rs6000-c: The return type of the following
21512 built-in functions was implemented as int not long long. Fix sign
21513 of return value for the unsigned version of vec_mulo and vec_mule.
21514 vector unsigned long long vec_bperm (vector unsigned long long,
21515 vector unsigned char)
21516 vector signed long long vec_mule (vector signed int,
21518 vector unsigned long long vec_mule (vector unsigned int,
21519 vector unsigned int)
21520 vector signed long long vec_mulo (vector signed int,
21522 vector unsigned long long vec_mulo (vector unsigned int,
21523 vector unsigned int)
21524 * doc/extend.texi: Fix the documentation for the built-in
21527 2017-06-07 Carl Love <cel@us.ibm.com>
21530 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21533 2017-06-07 Carl Love <cel@us.ibm.com>
21535 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21536 support, Generate doublehv for signed int/float for BE case only.
21538 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21540 * doc/invoke.texi (mcx16): Rewrite.
21542 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21544 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21545 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21546 *mov<mode>_softfloat, and an anonymous splitter): Use
21547 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21549 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21551 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21553 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21554 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21555 (rs6000_debug_reg_global): Adjust.
21556 (rs6000_init_hard_regno_mode_ok): Adjust.
21557 (rs6000_dbx_register_number): Adjust.
21558 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
21559 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21560 Remove SPE_ACC and SPEFSCR.
21561 (REG_ALLOC_ORDER): Ditto.
21562 (FRAME_POINTER_REGNUM): Change to 111.
21563 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
21564 (REG_CLASS_NAMES): Ditto.
21565 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
21566 (REGISTER_NAMES): Ditto.
21567 (ADDITIONAL_REG_NAMES): Ditto.
21568 (rs6000_reg_names): Ditto.
21569 * config/rs6000/rs6000.md: Renumber some register number
21572 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21574 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
21576 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21577 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
21579 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
21580 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21581 Delete the SPE high registers.
21582 (REG_ALLOC_ORDER): Ditto.
21583 (enum reg_class): Remove SPE_HIGH_REGS.
21584 (REG_CLASS_NAMES): Ditto.
21585 (REG_CLASS_CONTENTS): Delete the SPE high registers.
21586 (REGISTER_NAMES): Ditto.
21587 (rs6000_reg_names): Ditto.
21588 * doc/tm.texi.in: Remove SPE as example.
21589 * doc/tm.texi: Regenerate.
21591 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21593 * config/rs6000/8540.md (ppc8540_brinc): Delete.
21594 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
21595 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
21596 * config/rs6000/rs6000.md (type): Remove "brinc".
21598 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21600 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
21601 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
21602 * config/rs6000/linuxspe.h: Delete file.
21603 * config/rs6000/rs6000.md: Don't include spe.md.
21604 * config/rs6000/spe.h: Delete file.
21605 * config/rs6000/spe.md: Delete file.
21606 * config/rs6000/t-rs6000: Remove spe.md.
21608 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21610 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
21611 (reg_or_none500mem_operand): Delete.
21612 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
21613 instead of reg_or_none500mem_operand.
21615 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21617 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21618 handling of SPE flags.
21619 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
21621 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21623 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21625 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
21626 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
21627 paired_divv2sf3): Similar.
21628 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
21629 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
21630 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
21632 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
21633 Rename the paired_* instruction patterns.
21634 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
21636 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
21637 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
21638 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
21639 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
21640 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
21641 PAIRED_VECTOR_MODE.
21642 (struct machine_function): Delete field spe_insn_chain_scanned_p.
21643 (spe_func_has_64bit_regs_p): Delete.
21644 (spe_expand_predicate_builtin): Delete.
21645 (spe_expand_evsel_builtin): Delete.
21646 (TARGET_DWARF_REGISTER_SPAN): Do not define.
21647 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
21648 (invalid_e500_subreg): Delete.
21649 (rs6000_legitimize_address): Always force_reg op2 as well, for
21650 paired single memory accesses.
21651 (rs6000_member_type_forces_blk): Delete.
21652 (rs6000_spe_function_arg): Delete.
21653 (rs6000_expand_unop_builtin): Delete SPE handling.
21654 (rs6000_expand_binop_builtin): Ditto.
21655 (spe_expand_stv_builtin): Delete.
21656 (bdesc_2arg_spe): Delete.
21657 (spe_expand_builtin): Delete.
21658 (spe_expand_predicate_builtin): Delete.
21659 (spe_expand_evsel_builtin): Delete.
21660 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
21661 (spe_init_builtins): Delete.
21662 (spe_func_has_64bit_regs_p): Delete.
21663 (savres_routine_name): Delete "info" parameter. Adjust callers.
21664 (rs6000_emit_stack_reset): Ditto.
21665 (rs6000_dwarf_register_span): Delete.
21666 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
21667 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
21668 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
21670 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
21672 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
21673 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
21674 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
21675 mulv2sf3, divv2sf3): Delete expanders.
21677 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21679 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
21681 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21683 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
21684 * config/rs6000/rs6000.c: Ditto.
21686 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21688 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
21689 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
21690 comparison_operator.
21692 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21694 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
21695 * config/rs6000/rs6000.opt: Ditto.
21696 * config/rs6000/t-rtems: Ditto.
21698 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21700 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
21701 TARGET_E500_SINGLE by 0, simplify.
21702 * config/rs6000/rs6000.c: Ditto.
21703 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
21704 (spe_build_register_parallel): Delete.
21705 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
21706 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
21707 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
21708 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
21709 (E500_CONVERT): Delete.
21710 * config/rs6000/spe.md: Remove many patterns and all define_constants.
21712 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21714 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
21715 * config/rs6000/dfp.md: Ditto.
21716 (negdd2, *negdd2_fpr): Merge.
21717 (absdd2, *absdd2_fpr): Merge.
21718 (negtd2, *negtd2_fpr): Merge.
21719 (abstd2, *abstd2_fpr): Merge.
21720 * config/rs6000/e500.h: Delete file.
21721 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
21722 TARGET_FPRS by 1 and simplify.
21723 * config/rs6000/rs6000-c.c: Ditto.
21724 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
21725 TARGET_DF_SPE by 0.
21726 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
21728 * config/rs6000/rs6000.md: Ditto.
21729 (floatdidf2, *floatdidf2_fpr): Merge.
21730 (move_from_CR_gt_bit): Delete.
21731 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
21732 (E500_CR_IOR_COMPARE): Delete.
21733 (All patterns that require !TARGET_FPRS): Delete.
21734 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
21736 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21738 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21740 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21742 * graphds.c (add_edge): Intitialize edge's attached data.
21743 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21744 pointer parameter. Call pointed function on each edge during
21745 graph traversing. Skip traversing the edge when the function
21747 (graphds_dfs, graphds_scc): Ditto.
21748 (for_each_edge): New parameter. Pass the new parameter to callback
21750 * graphds.h (skip_edge_callback): New function pointer type.
21751 (graphds_dfs, graphds_scc): New function pointer parameter.
21752 (graphds_edge_callback, for_each_edge): New parameter.
21754 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21756 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21757 out code checking if runtime alias check is possible to below ...
21758 Call the new function.
21759 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21760 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21762 2017-06-07 Marek Polacek <polacek@redhat.com>
21765 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21766 TYPE_OVERFLOW_WRAPS check.
21768 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21770 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
21771 if versioning is required.
21772 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
21773 peeling with the check for versioning.
21775 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21777 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
21778 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
21779 Set true to new parameter if new ssa variable is defined.
21780 (vect_gen_vector_loop_niters): Refactor. Set range information
21781 for the new vector loop bound variable.
21782 (vect_do_peeling): Ditto.
21784 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21786 * tree-affine.c (ssa.h): Include header file.
21787 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
21788 has wrapping overflow behavior.
21790 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21792 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
21794 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21796 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
21797 (tree_to_aff_combination): ... here.
21799 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21801 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
21802 reg_pressure model function.
21803 (ivopts_global_cost_for_size): Delete.
21804 (determine_set_costs, iv_ca_recount_cost): Call new model function
21805 ivopts_estimate_reg_pressure.
21807 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21809 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
21810 expensive than udiv. Remove floating point cases from mod.
21812 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21814 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
21815 Increase idiv cost.
21817 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21819 * config/aarch64/aarch64.md
21820 (copysignsf3): Fix mask generation.
21822 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21824 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21826 (class dump_manager): Add register_dumps method.
21827 * dumpfile.c: Include langhooks.h.
21828 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21829 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21830 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21831 (dump_manager::dump_register): Start with 512 entries instead of 32.
21832 (dump_manager::register_dumps): New method.
21833 * toplev.c (general_init): Instead of invoking register_dumps
21834 langhook, invoke register_dumps method on the dump manager.
21835 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21838 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21840 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21841 Say that # requires an associated define_split to exist, and that
21842 the define_split must be suitable for use after register allocation.
21844 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21846 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21847 (compute_outgoing_frequencies): Also initialize zero counts.
21848 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21849 around loops; preserve more of profile when nothing changes.
21851 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21853 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21855 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21856 * config/arm/arm-cpu-cdata.h: Regenerate.
21857 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21858 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21859 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21860 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21861 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21863 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21865 * config/arm/t-rmprofile: Likewise.
21866 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21868 2017-06-06 David S. Miller <davem@davemloft.net>
21871 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21872 blockage if function uses alloca.
21874 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21876 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21877 New "uid" fields to hold pretty-print IDs of group and ref.
21878 Memory references are now identified as <group_id>:<ref_id>
21879 instead of using [random] addresses.
21880 (dump_mem_details): Simplify, no functional change.
21881 (dump_mem_ref): Simplify and make output more concise.
21882 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21883 (find_or_create_group): Initialize group uid.
21884 (record_ref): Initialize ref uid. Improve debug output.
21885 (prune_group_by_reuse, should_issue_prefetch_p,)
21886 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21887 (mark_nontemporal_store, determine_loop_nest_reuse):
21888 Improve debug output.
21890 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21892 * dbgcnt.def (prefetch): New debug counter.
21893 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21894 (schedule_prefetches): Stop issueing prefetches if debug counter
21897 2017-06-06 Tom de Vries <tom@codesourcery.com>
21899 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21900 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21902 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21904 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21905 Use aarch64_reg_or_zero predicate for operand 4.
21906 (aarch64_compare_and_swap<mode> define_insn_and_split):
21907 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21908 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21910 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21912 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21913 (arm_compute_save_core_reg_mask): This.
21914 (thumb1_compute_save_reg_mask): Rename into ...
21915 (thumb1_compute_save_core_reg_mask): This.
21916 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21917 (arm_compute_frame_layout): Likewise.
21919 2017-06-06 Richard Biener <rguenther@suse.de>
21921 PR tree-optimization/80974
21922 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21923 keep or clear leaders SSA info.
21925 2017-06-06 Tom de Vries <tom@codesourcery.com>
21927 * config/nvptx/nvptx.c (split_mode_p): New function.
21928 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21930 2017-06-06 Tom de Vries <tom@codesourcery.com>
21932 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21934 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21937 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21940 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21942 * shrink-wrap.c (handle_simple_exit): Update profile.
21943 (try_shrink_wrapping): Upate profile.
21945 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21947 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21948 (tree_guess_outgoing_edge_probabilities): New.
21949 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21950 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21952 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21954 * ipa-split.c (split_function): Initialize return bb profile.
21956 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21958 * profile.c (compute_branch_probabilities): Also initialize
21959 EXIT_BLOCK profile.
21961 2017-06-06 Richard Biener <rguenther@suse.de>
21963 PR tree-optimization/80928
21964 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21965 (vect_analyze_loop_operations): Properly guard analysis for
21967 (vect_transform_loop): Likewise.
21968 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21969 (vect_model_induction_cost): Do not cost for pure SLP.
21970 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21971 of induction in inner loop vectorization.
21972 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21973 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21974 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21976 (vect_analyze_slp_cost_1): Cost induction.
21977 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21978 (vect_get_slp_vect_defs): Likewise.
21979 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21980 (vect_transform_stmt): Handle SLP reductions.
21981 * tree-vectorizer.h (vectorizable_induction): Adjust.
21983 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21985 * config/rs6000/rs6000.c (make_resolver_func): Update
21986 init_lowered_empty_function call.
21988 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21990 * doc/invoke.texi: Document the -fprofile-abs-path option.
21991 * common.opt (fprofile-abs-path): New option.
21992 * gcov-io.h (gcov_write_filename): Declare.
21993 * gcov-io.c (gcov_write_filename): New function.
21994 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21995 * profile.c (output_location): Likewise.
21997 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21999 * shring-wrap.c: Revert accidental commit.
22001 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22003 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22005 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22007 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22009 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22010 profile in callgraph edge.
22011 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22012 (apply_scale): Likewise.
22013 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22014 Also scale profile when inlining function with zero profile.
22015 (initialize_cfun): Update exit block profile even when it is zero.
22016 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22017 when profile is read.
22019 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22021 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22022 (CLONE_*): New constants to define the processors we can generate
22023 code for with the target_clone attribute.
22024 (rs6000_clone_map): New array to identify which clone processors
22025 the current program is running on.
22026 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22027 target_clone attribute.
22028 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22029 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22030 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22031 (cpu_expand_builtin): Add support for target_clone attribute.
22032 (rs6000_valid_attribute_p): Allow "default" attribute.
22033 (get_decl_name): New debug function to simplify printing the
22034 current function name in debugging statements.
22035 (rs6000_clone_priority): New functions to support the target_clone
22036 attribute, and be able to generate code to switch between ISA 2.05
22037 through ISA 3.0 (power6 through power9).
22038 (rs6000_compare_version_priority): Likewise.
22039 (rs6000_get_function_versions_dispatcher): Likewise.
22040 (make_resolver_func): Likewise.
22041 (add_condition_to_bb): Likewise.
22042 (dispatch_function_versions): Likewise.
22043 (rs6000_generate_version_dispatcher_body): Likewise.
22044 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22045 (fusion_gpr_load_p): Fix a spacing issue.
22046 * doc/extend.texi (Common Function Attributes): Document that the
22047 PowerPC supports the target_clone attribute.
22049 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22051 * config/arm/arm.h: explain F symbol found in description of ARM
22052 register allocation in its legend.
22054 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22056 * config/mips/frame-header-opt.c: Include profile-count.h.
22057 * config/riscv/riscv.c: Include profile-count.h
22059 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22061 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22063 (sm_set_flag_if_changed): Add bbs field.
22064 (execute_sm_if_changed_flag_set): Pass BBS.
22065 (execute_sm): Update.
22067 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22069 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22072 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22074 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22075 (peephole2): New peephole2 to emit the above.
22076 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22078 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22080 * config/aarch64/aarch64.c (define_peephole2 above
22081 *sub_<shift>_<mode>): New peephole.
22083 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22085 * config/i386/i386.c (make_resolver_func): Update.
22086 * Makefile.in: Add profile-count.h and profile-count.o
22087 * auto-profile.c (afdo_indirect_call): Update to new API.
22088 (afdo_set_bb_count): Update.
22089 (afdo_propagate_edge): Update.
22090 (afdo_propagate_circuit): Update.
22091 (afdo_calculate_branch_prob): Update.
22092 (afdo_annotate_cfg): Update.
22093 * basic-block.h: Include profile-count.h
22094 (struct edge_def): Turn count to profile_count.
22095 (struct basic_block_def): Likewie.
22096 (REG_BR_PROB_BASE): Move to profile-count.h
22097 (RDIV): Move to profile-count.h
22098 * bb-reorder.c (max_entry_count): Turn to profile_count.
22099 (find_traces): Update.
22100 (rotate_loop):Update.
22101 (connect_traces):Update.
22102 (sanitize_hot_paths):Update.
22103 * bt-load.c (migrate_btr_defs): Update.
22104 * cfg.c (RDIV): Remove.
22105 (init_flow): Use alloc_block.
22106 (alloc_block): Uninitialize count.
22107 (unchecked_make_edge): Uninitialize count.
22108 (check_bb_profile): Update.
22109 (dump_edge_info): Update.
22110 (dump_bb_info): Update.
22111 (update_bb_profile_for_threading): Update.
22112 (scale_bbs_frequencies_int): Update.
22113 (scale_bbs_frequencies_gcov_type): Update.
22114 (scale_bbs_frequencies_profile_count): New.
22115 * cfg.h (update_bb_profile_for_threading): Update.
22116 (scale_bbs_frequencies_profile_count): Declare.
22117 * cfgbuild.c (compute_outgoing_frequencies): Update.
22118 (find_many_sub_basic_blocks): Update.
22119 * cfgcleanup.c (try_forward_edges): Update.
22120 (try_crossjump_to_edge): Update.
22121 * cfgexpand.c (expand_gimple_tailcall): Update.
22122 (construct_exit_block): Update.
22123 * cfghooks.c (verify_flow_info): Update.
22124 (dump_bb_for_graph): Update.
22125 (split_edge): Update.
22126 (make_forwarder_block): Update.
22127 (duplicate_block): Update.
22128 (account_profile_record): Update.
22129 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22130 (get_estimated_loop_iterations): Update.
22131 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22132 (single_likely_exit): Update.
22133 * cfgloopmanip.c (scale_loop_profile): Update.
22135 (set_zero_probability): Update.
22136 (lv_adjust_loop_entry_edge): Update.
22137 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22138 (purge_dead_edges): Update.
22139 (rtl_account_profile_record): Update.
22140 * cgraph.c (cgraph_node::create): Uninitialize count.
22141 (symbol_table::create_edge): Uninitialize count.
22142 (cgraph_update_edges_for_call_stmt_node): Update.
22143 (cgraph_edge::dump_edge_flags): Update.
22144 (cgraph_node::dump): Update.
22145 (cgraph_edge::maybe_hot_p): Update.
22146 * cgraph.h: Include profile-count.h
22147 (create_clone), create_edge, create_indirect_edge): Update.
22148 (cgraph_node): Turn count to profile_count.
22149 (cgraph_edge0: Likewise.
22150 (make_speculative, clone): Update.
22151 (create_edge): Update.
22152 (init_lowered_empty_function): Update.
22153 * cgraphclones.c (cgraph_edge::clone): Update.
22154 (duplicate_thunk_for_node): Update.
22155 (cgraph_node::create_clone): Update.
22156 * cgraphunit.c (cgraph_node::analyze): Update.
22157 (cgraph_node::expand_thunk): Update.
22158 * final.c (dump_basic_block_info): Update.
22159 * gimple-streamer-in.c (input_bb): Update.
22160 * gimple-streamer-out.c (output_bb): Update.
22161 * graphite.c (print_global_statistics): Update.
22162 (print_graphite_scop_statistics): Update.
22163 * hsa-brig.c: Include basic-block.h.
22164 * hsa-dump.c: Include basic-block.h.
22165 * hsa-gen.c (T sum_slice): Update.
22166 (convert_switch_statements):Update.
22167 * hsa-regalloc.c: Include basic-block.h.
22168 * ipa-chkp.c (chkp_produce_thunks): Update.
22169 * ipa-cp.c (struct caller_statistics): Update.
22170 (init_caller_stats): Update.
22171 (gather_caller_stats): Update.
22172 (ipcp_cloning_candidate_p): Update.
22173 (good_cloning_opportunity_p): Update.
22174 (get_info_about_necessary_edges): Update.
22175 (dump_profile_updates): Update.
22176 (update_profiling_info): Update.
22177 (update_specialized_profile): Update.
22178 (perhaps_add_new_callers): Update.
22179 (decide_about_value): Update.
22180 (ipa_cp_c_finalize): Update.
22181 * ipa-devirt.c (struct odr_type_warn_count): Update.
22182 (struct decl_warn_count): Update.
22183 (struct final_warning_record): Update.
22184 (possible_polymorphic_call_targets): Update.
22185 (ipa_devirt): Update.
22186 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22187 * ipa-icf.c (sem_function::merge): Update.
22188 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22189 * ipa-inline.c (compute_uninlined_call_time): Update.
22190 (compute_inlined_call_time): Update.
22191 (want_inline_small_function_p): Update.
22192 (want_inline_self_recursive_call_p): Update.
22193 (edge_badness): Update.
22194 (lookup_recursive_calls): Update.
22195 (recursive_inlining): Update.
22196 (inline_small_functions): Update.
22197 (dump_overall_stats): Update.
22198 (dump_inline_stats): Update.
22199 * ipa-profile.c (ipa_profile_generate_summary): Update.
22200 (ipa_propagate_frequency): Update.
22201 (ipa_profile): Update.
22202 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22203 * ipa-utils.c (ipa_merge_profiles): Update.
22204 * loop-doloop.c (doloop_modify): Update.
22205 * loop-unroll.c (report_unroll): Update.
22206 (unroll_loop_runtime_iterations): Update.
22207 * lto-cgraph.c (lto_output_edge): Update.
22208 (lto_output_node): Update.
22209 (input_node): Update.
22210 (input_edge): Update.
22211 (merge_profile_summaries): Update.
22212 * lto-streamer-in.c (input_cfg): Update.
22213 * lto-streamer-out.c (output_cfg): Update.
22214 * mcf.c (create_fixup_graph): Update.
22215 (adjust_cfg_counts): Update.
22216 (sum_edge_counts): Update.
22217 * modulo-sched.c (sms_schedule): Update.
22218 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22219 * predict.c (maybe_hot_count_p): Update.
22220 (probably_never_executed): Update.
22221 (dump_prediction): Update.
22222 (combine_predictions_for_bb): Update.
22223 (propagate_freq): Update.
22224 (handle_missing_profiles): Update.
22225 (counts_to_freqs): Update.
22226 (rebuild_frequencies): Update.
22227 (force_edge_cold): Update.
22228 * predict.h: Include profile-count.h
22229 (maybe_hot_count_p, counts_to_freqs): UPdate.
22230 * print-rtl-function.c: Do not include cfg.h
22231 * print-rtl.c: Include basic-block.h
22232 * profile-count.c: New file.
22233 * profile-count.h: New file.
22234 * profile.c (is_edge_inconsistent): Update.
22235 (correct_negative_edge_counts): Update.
22236 (is_inconsistent): Update.
22237 (set_bb_counts): Update.
22238 (read_profile_edge_counts): Update.
22239 (compute_frequency_overlap): Update.
22240 (compute_branch_probabilities): Update; Initialize and deinitialize
22242 (branch_prob): Update.
22243 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22244 (edge_gcov_count): New.
22245 (bb_gcov_count): New.
22246 * shrink-wrap.c (try_shrink_wrapping): Update.
22247 * tracer.c (better_p): Update.
22248 * trans-mem.c (expand_transaction): Update.
22249 (ipa_tm_insert_irr_call): Update.
22250 (ipa_tm_insert_gettmclone_call): Update.
22251 * tree-call-cdce.c: Update.
22252 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22253 (gimple_duplicate_sese_tail): Update.
22254 (gimple_account_profile_record): Update.
22255 (execute_fixup_cfg): Update.
22256 * tree-inline.c (copy_bb): Update.
22257 (copy_edges_for_bb): Update.
22258 (initialize_cfun): Update.
22259 (freqs_to_counts): Update.
22260 (copy_cfg_body): Update.
22261 (expand_call_inline): Update.
22262 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22263 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22264 (try_unroll_loop_completely): Update.
22265 (try_peel_loop): Update.
22266 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22267 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22268 * tree-ssa-loop-split.c (connect_loops): Update.
22269 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22270 * tree-ssa-reassoc.c (branch_fixup): Update.
22271 * tree-ssa-tail-merge.c (replace_block_by): Update.
22272 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22273 (compute_path_counts): Update.
22274 (update_profile): Update.
22275 (recompute_probabilities): Update.
22276 (update_joiner_offpath_counts): Update.
22277 (estimated_freqs_path): Update.
22278 (freqs_to_counts_path): Update.
22279 (clear_counts_path): Update.
22280 (ssa_fix_duplicate_block_edges): Update.
22281 (duplicate_thread_path): Update.
22282 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22283 (struct switch_conv_info): Update.
22284 * tree-tailcall.c (decrease_profile): Update.
22285 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22286 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22287 * value-prof.c (check_counter): Update.
22288 (gimple_divmod_fixed_value): Update.
22289 (gimple_mod_pow2): Update.
22290 (gimple_mod_subtract): Update.
22291 (gimple_ic_transform): Update.
22292 (gimple_stringop_fixed_value): Update.
22293 * value-prof.h (gimple_ic): Update.
22295 2017-06-02 Carl Love <cel@us.ibm.com>
22297 * config/rs6000/rs6000-c: Add support for built-in functions
22298 vector double vec_doublee (vector signed int);
22299 vector double vec_doublee (vector unsigned int);
22300 vector double vec_doublee (vector float);
22301 vector double vec_doubleh (vector signed int);
22302 vector double vec_doubleh (vector unsigned int);
22303 vector double vec_doubleh (vector float);
22304 vector double vec_doublel (vector signed int);
22305 vector double vec_doublel (vector unsigned int);
22306 vector double vec_doublel (vector float);
22307 vector double vec_doubleo (vector signed int);
22308 vector double vec_doubleo (vector unsigned int);
22309 vector double vec_doubleo (vector float);.
22310 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22311 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22313 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22314 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22315 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22317 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22318 vec_doublel, vec_doubleh.
22319 * doc/extend.texi: Update the built-in documentation file for the
22320 new built-in functions.
22322 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22325 * ipa-inline-analysis.c (free_growth_caches): Set
22326 edge_removal_hook_holder to NULL after removing it.
22328 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22330 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22331 comparision with zero.
22333 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22334 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22335 for early expansion of vec_min and vec_max builtins.
22336 (builtin_function_type): Add min/max unsigned variants to those
22337 identified as having unsigned arguments.
22339 2017-06-02 Olivier Hainque <hainque@adacore.com>
22341 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22343 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22345 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22346 Use VALL_F16 iterator rather than VALL.
22348 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22350 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22351 Emit CBNZ inside loop when doing a strong exchange and comparing
22352 against zero. Generate the CC flags after the loop.
22354 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22356 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22357 (dl_section_ref): New.
22358 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22359 On AIX, append an expression to subtract the size of the
22360 section length to dl_section_ref.
22362 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22364 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22365 for early expansion of vector absolute builtins.
22367 2017-06-02 Richard Biener <rguenther@suse.de>
22369 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22370 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22372 2017-06-02 Richard Biener <rguenther@suse.de>
22374 PR tree-optimization/80948
22375 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22376 stmt order as well.
22378 2017-06-02 Richard Biener <rguenther@suse.de>
22380 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22382 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22383 for inductions as relevant.
22385 2017-06-02 Richard Biener <rguenther@suse.de>
22387 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22388 (vectorizable_induction): ... this. Remove dead code.
22390 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22392 * builtins. (expand_builtin_alloca): Remove second parameter and
22393 infer its value from the first parameter instead.
22394 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22396 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22398 PR rtl-optimization/80903
22399 * loop-doloop.c (add_test): Unshare sequence.
22401 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22403 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22405 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22407 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22409 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22410 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22411 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22412 (xlogue_layout::get_stub_rtx): Make static.
22413 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22414 (xlogue_layout::compute_stub_managed_regs): Rename to...
22415 (xlogue_layout::count_stub_managed_regs): ...this.
22416 (xlogue_layout::is_stub_managed_reg): New function.
22417 (xlogue_layout::m_stub_names): Rename to...
22418 (xlogue_layout::s_stub_names): ...this, make static.
22419 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22420 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22421 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22422 xlogue_layout::s_stub_names): Instantiate statics.
22423 (stub_managed_regs): Remove.
22424 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22425 (disable_call_ms2sysv_xlogues): Rename to...
22426 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22427 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22429 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22430 change after reload_completed.
22431 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22433 (ix86_expand_prologue): Likewise.
22434 (ix86_expand_epilogue): Likewise.
22435 (ix86_expand_split_stack_prologue): Likewise.
22436 (ix86_compute_frame_layout): Remove frame parameter ...
22437 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22438 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22440 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22441 (ix86_frame): Move from here ...
22442 * config/i386/i386.h (ix86_frame): ... to here.
22443 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22444 complete ix86_frame data structure instead. Remove some_ld_name.
22446 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22448 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22449 symbols that hold a DECL_VALUE_EXPR.
22451 2017-06-01 Martin Jambor <mjambor@suse.cz>
22453 PR tree-optimization/80898
22454 * tree-sra.c (process_subtree_disqualification): Removed.
22455 (disqualify_candidate): Do not acll
22456 process_subtree_disqualification.
22457 (subtree_mark_written_and_enqueue): New function.
22458 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22459 RHS has been disqualified and re-queue LHS if necessary. Apart
22460 from that, ignore disqualified RHS.
22462 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22464 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22465 address fetch for z10 or later.
22467 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22469 * config/arc/arc.md (tst_movb): Add guard when splitting.
22471 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22473 * config/arc/arc.c (arc_can_eliminate): Test against
22474 arc_frame_pointer_needed.
22476 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22478 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22479 to prevent store reordering.
22480 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22481 (type): Add block type.
22482 (stack_tie): Define special instruction to be used in
22485 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22487 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22488 constraint. It is not valid for the pattern.
22489 (noncommutative_binary_comparison): Likewise.
22491 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22493 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22496 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22498 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22499 be used by the reg-alloc.
22501 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22503 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22504 reg-alloc when having mul64 or mul32x16 instructions.
22505 (mulsidi3): Likewise.
22506 (umulsidi3): Likewise.
22507 (mulsi32x16): New pattern.
22508 (mulsi64): Likewise.
22509 (mulsidi64): Likewise.
22510 (umulsidi64): Likewise.
22511 (MUL32x16_REG): Define.
22512 (mul64_600): Use MUL32x16_REG.
22513 (mac64_600): Likewise.
22514 (umul64_600): Likewise.
22515 (umac64_600): Likewise.
22517 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22519 * config/arc/arc.md (mulsi3_700): Make it commutative.
22521 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22523 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22525 (*sign_extendsidi2_insn): Likewise for movstosw.
22527 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22529 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22530 the type of the input discriminant value. Convert the
22531 discriminant value of signedness vary.
22533 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22535 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22536 Add to -Wall section.
22538 2017-06-01 Richard Biener <rguenther@suse.de>
22540 PR middle-end/66313
22541 * fold-const.c (fold_plusminus_mult_expr): If the factored
22542 factor may be zero use a wrapping type for the inner operation.
22543 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22544 and handle moved defs.
22545 (process_assignment): Properly guard the unary op case. Return a
22546 tri-state indicating that moving the stmt before the call may allow
22547 to continue. Pass through to_move.
22548 (find_tail_calls): Handle moving unrelated defs before
22551 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22554 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22555 splitter result in the canonical way.
22557 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
22559 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
22560 also for 32bit target. Update insn attributes.
22561 (zero-extendsidi2 splitter): Allow all registers for operand 1.
22563 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
22565 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
22566 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
22567 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
22568 (_mm_maskz_min_ss): New intrinsics.
22570 2017-05-31 Martin Liska <mliska@suse.cz>
22572 * tree-vect-loop.c (vect_create_epilog_for_reduction):
22573 Change comment style to one we normally use.
22574 (vectorizable_reduction): Likewise.
22575 (vectorizable_induction): Likewise.
22576 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
22577 (vectorizable_call): Likewise.
22578 (vectorizable_simd_clone_call): Likewise.
22579 (vectorizable_conversion): Likewise.
22580 (vectorizable_assignment): Likewise.
22581 (vectorizable_shift): Likewise.
22582 (vectorizable_operation): Likewise.
22583 (vectorizable_store): Likewise.
22584 (vectorizable_load): Likewise.
22585 * tree-vectorizer.h: Likewise.
22587 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
22589 * passes.c (emergency_dump_function): New.
22590 * tree-pass.h (emergency_dump_function): Declare.
22591 * plugin.c (plugins_internal_error_function): Remove.
22592 * plugin.h (plugins_internal_error_function): Remove declaration.
22593 * toplev.c (internal_error_function): New static function. Use it...
22594 (general_init): ...here.
22596 2017-05-31 Graham Markall <graham.markall@embecosm.com>
22598 * config/arc/arc.c (arc_print_operand): Handle constant operands.
22599 (arc_rtx_costs): Add costs for new patterns.
22600 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
22601 * config/arc/predicates.md: Add _1_2_3_operand predicate.
22603 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22605 * tree-ssa-strlen.c (get_next_strinfo): New function.
22606 (get_stridx_plus_constant): Use it.
22607 (zero_length_string): Likewise.
22608 (adjust_related_strinfos): Likewise.
22609 (adjust_last_stmt): Likewise.
22611 2017-05-31 Richard Biener <rguenther@suse.de>
22614 * config/i386/i386.c (ix86_expand_builtin): Remove assert
22615 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
22617 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22619 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
22620 loop_vinfo argument and use of dependence distance vectors.
22621 Check instead whether the two references differ only in their
22622 initial value and assume that they have the same alignment if the
22623 difference is a multiple of the vector alignment.
22624 (vect_analyze_data_refs_alignment): Update call accordingly.
22626 2017-05-31 Martin Liska <mliska@suse.cz>
22629 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
22631 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22633 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
22634 (create_intersect_range_checks): Move from ...
22635 * tree-data-ref.c (create_intersect_range_checks_index)
22636 (create_intersect_range_checks): ... to here.
22637 (create_runtime_alias_checks): New function factored from ...
22638 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
22639 here. Call above function.
22640 * tree-data-ref.h (create_runtime_alias_checks): New function.
22642 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22644 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
22645 segment length for dr_b and compute it in wide_int.
22647 2017-05-31 Richard Biener <rguenther@suse.de>
22649 PR tree-optimization/80906
22650 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
22651 and pass through iv_map.
22652 (copy_bb_and_scalar_dependences): Adjust.
22653 (translate_pending_phi_nodes): Likewise.
22654 (copy_loop_close_phi_args): Handle code-generating IVs instead
22657 2017-05-30 David Malcolm <dmalcolm@redhat.com>
22659 * diagnostic-color.c (color_dict): Add "type-diff".
22660 (parse_gcc_colors): Update comment.
22661 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
22662 -fdiagnostics-show-template-tree and -fno-elide-type.
22663 (GCC_COLORS): Add type-diff to example.
22665 (-fdiagnostics-show-template-tree): New.
22666 (-fno-elide-type): New.
22667 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
22668 the pp_format_decoder callback. Call any m_format_postprocessor's
22670 (pretty_printer::pretty_printer): Initialize
22671 m_format_postprocessor.
22672 (pretty_printer::~pretty_printer): Delete any
22673 m_format_postprocessor.
22674 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
22675 (class format_postprocessor): New class.
22676 (struct pretty_printer::format_decoder): Document the new parameters.
22677 (struct pretty_printer::m_format_postprocessor): New field.
22678 * tree-diagnostic.c (default_tree_printer): Update for new
22679 bool and const char ** params.
22680 * tree-diagnostic.h (default_tree_printer): Likewise.
22682 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
22684 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
22685 (lwa_operand): Delete rs6000_gen_cell_microcode test.
22686 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22687 rs6000_gen_cell_microcode code.
22688 (rs6000_final_prescan_insn): Delete.
22689 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
22690 "warn-cell-microcode" entries.
22691 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
22692 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
22693 throughout. Change cc_reg_not_micro_cr0_operand to
22694 cc_reg_not_cr0_operand throughout.
22695 (*extendhi<mode>2_noload): Delete.
22696 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
22697 (mwarn-cell-microcode): Delete.
22698 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
22699 -mgen-cell-microcode and -mwarn-cell-microcode.
22701 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
22704 * config/i386/constraints.md (Yd): New constraint.
22706 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
22707 and (?Yd, r) alternatives. Update insn attributes.
22708 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
22709 and (?*Yd, r) alternatives. Update insn attributes.
22710 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
22712 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
22714 * gimplify.c (gimplify_modify_expr): Don't create a
22715 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
22718 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
22720 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
22722 2017-05-30 Richard Biener <rguenther@suse.de>
22724 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
22725 and reduc_def fields.
22726 (STMT_VINFO_REDUC_TYPE): New define.
22727 (STMT_VINFO_REDUC_DEF): Likewise.
22728 (vect_force_simple_reduction): Adjust prototype.
22729 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
22730 (vect_is_simple_reduction): Remove check_reduction argument.
22731 (vect_force_simple_reduction): Adjust and set
22732 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22733 (vectorizable_reduction): Do not re-do reduction analysis
22734 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22735 * tree-parloops.c (gather_scalar_reductions): Adjust.
22737 2017-05-30 Richard Biener <rguenther@suse.de>
22739 PR middle-end/80901
22740 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22743 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22745 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22746 Introduce unknown_misalignment parameter and remove vf.
22747 (vect_peeling_hash_get_lowest_cost):
22748 Pass unknown_misalignment parameter.
22749 (vect_enhance_data_refs_alignment):
22750 Fix unsupportable data ref treatment.
22752 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22754 * tree-vect-data-refs.c (vect_get_data_access_cost):
22755 Workaround for SLP handling.
22756 (vect_enhance_data_refs_alignment):
22757 Compute costs for doing no peeling at all, compare to the best
22758 peeling costs so far and avoid peeling if cheaper.
22760 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22762 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22763 Return peeling info and set costs to zero for unlimited cost
22765 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22766 with unknown misalignment. Compute and costs for unknown
22767 misalignment, compare them to the costs for known misalignment
22768 and choose the cheapest for peeling.
22770 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22772 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
22773 (vect_get_peeling_costs_all_drs): Create function.
22774 (vect_peeling_hash_get_lowest_cost):
22775 Use vect_get_peeling_costs_all_drs.
22776 (vect_peeling_supportable): Create function.
22777 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
22779 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22781 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
22782 DR_HAS_NEGATIVE_STEP.
22783 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
22784 (vect_enhance_data_refs_alignment): Use.
22785 (vect_duplicate_ssa_name_ptr_info): Use.
22786 * tree-vectorizer.h (dr_misalignment): Use.
22787 (known_alignment_for_access_p): Use.
22789 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22792 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
22794 (has_section_name): New function.
22796 2017-05-30 Martin Liska <mliska@suse.cz>
22799 * auto-profile.c (get_function_decl_from_block): Fix
22802 2017-05-30 Richard Biener <rguenther@suse.de>
22804 PR middle-end/80876
22805 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
22807 2017-05-30 Martin Liska <mliska@suse.cz>
22809 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
22810 * dumpfile.h (struct dump_file_info): Remove ctors.
22812 2017-05-30 Martin Liska <mliska@suse.cz>
22814 * predict.def: Fix GNU coding style.
22816 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22818 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22819 Mark 'to' argument with ATTRIBUTE_UNUSED.
22821 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22823 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22824 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22825 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22828 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22830 * doc/install.texi (Options specification): Restore entry of
22831 --enable-sjlj-exceptions.
22833 2017-05-27 Michael Eager <eager@eagercon.com>
22836 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22838 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22840 * config/microblaze/microblaze.h
22841 (FIXED_REGISTERS): Update in macro.
22842 (CALL_USED_REGISTERS): Update in macro.
22844 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22846 * doc/install.texi: Add links to macOS binary distributions.
22848 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22852 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22854 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22856 2017-05-26 Martin Liska <mliska@suse.cz>
22858 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22860 2017-05-26 Martin Liska <mliska@suse.cz>
22862 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22863 always leading ';; '.
22864 (dump_bb_info): Likewise.
22865 (brief_dump_cfg): Likewise.
22866 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22867 * dumpfile.c: Remove usage of TDF_VERBOSE.
22868 * dumpfile.h (enum dump_kind): Likewise.
22869 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22870 * print-tree.c (debug_verbose): Remove.
22871 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22872 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22873 * tree-diagnostic.c (default_tree_printer): Replace
22874 TDF_DIAGNOSTIC with TDF_SLIM.
22876 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22878 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22879 in parameter loop, rather than loop_vinfo.
22880 (create_intersect_range_checks): Ditto.
22881 (vect_create_cond_for_alias_checks): Update call to above functions.
22883 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22885 PR tree-optimization/80815
22886 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22887 for merging runtime alias checks. Handle negative DR_STEPs.
22889 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22891 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22893 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22894 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22895 out code pruning runtime alias checks.
22896 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22897 factored out from above.
22898 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22900 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22902 (prune_runtime_alias_test_list): New decalaration.
22904 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22906 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22907 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22908 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22909 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22910 (operator==, comp_dr_with_seg_len_pair): Ditto.
22911 (vect_prune_runtime_alias_test_list): Ditto.
22913 2017-05-26 Martin Liska <mliska@suse.cz>
22916 * params.def: Bound partial-inlining-entry-probability param.
22918 2017-05-26 Marek Polacek <polacek@redhat.com>
22921 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22924 2017-05-26 Richard Biener <rguenther@suse.de>
22926 PR tree-optimization/80842
22927 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22930 2017-05-26 Richard Biener <rguenther@suse.de>
22932 PR tree-optimization/80844
22933 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22935 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22937 * doc/md.texi (Machine Constraints): Update x86 family
22938 machine constraints section to match 'config/i386/constraints.md'.
22940 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22942 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22944 2017-05-25 Nathan Sidwell <nathan@acm.org>
22946 * doc/invoke.texi (--enable-languages): Update documentation.
22948 2017-05-25 Martin Liska <mliska@suse.cz>
22950 * dumpfile.c: Add TDF_FOLDING.
22951 * dumpfile.h (enum dump_kind): Likewise.
22952 * genmatch.c (dt_simplify::gen_1): Use it.
22954 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22956 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22958 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22960 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22961 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22963 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22965 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22966 * match.pd (X == C): Rewrite it here.
22967 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22968 with_certain_nonzero_bits2): New predicates.
22969 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22971 2017-05-24 Nathan Sidwell <nathan@acm.org>
22973 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22976 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22979 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22981 * config/powerpcspe: New port. Files are copied from the rs6000
22982 port, with "rs6000" in filenames replaced by "powerpcspe".
22984 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22986 PR rtl-optimization/80754
22987 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22989 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22991 * config/sparc/sparc.md (length): Return the correct value for -mflat
22992 sibcalls to match output_sibcall.
22994 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22998 * config/rs6000/rs6000.c (struct machine_function): Add new field
23000 (rs6000_get_separate_components): Init that field, use it.
23001 (rs6000_components_for_bb): Use the field.
23003 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23005 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23007 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23009 PR middle-end/80823
23010 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23012 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23015 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23016 address against address_operand predicate.
23017 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23019 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23021 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23022 parameters passed indirectly.
23024 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23026 * config/i386/i386.md (*movdi_internal): Remove SSE4
23027 alternative 18 (?r, *v). Update insn attributes.
23028 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23029 Update insn attributes.
23030 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23031 Update insn attributes.
23032 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23033 alternative 1 (r, v). Remove isa attribute.
23034 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23035 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23036 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23038 2017-05-23 Tom de Vries <tom@codesourcery.com>
23040 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23043 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23045 * cgraphunit.c (symbol_table::process_new_functions): Update.
23046 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23047 (inline_generate_summary): Rename to ...
23048 (ipa_fn_summary_generate): ... this one.
23049 (inline_read_summary): Rename to ...
23050 (ipa_fn_summary_read): ... this one.
23051 (inline_write_summary): Rename to ...
23052 (ipa_fn_summary_write): ... this one.
23053 (inline_free_summary): Rename to ...
23054 (ipa_free_fn_summary): ... this one.
23055 (pass_data_local_fn_summary, pass_local_fn_summary,
23056 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23057 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23058 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23059 make_pass_ipa_fn_summary): New.
23060 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23061 inline_write_summary, inline_free_summary): Remove.
23062 (ipa_free_fn_summary) : New.
23063 * ipa-inline.c (ipa_inline): Update.
23064 (pass_ipa_inline): Do not generate summaries.
23065 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23067 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23068 and add pass_ipa_fn_summary.
23069 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23071 (make_pass_inline_parameters): Remove.
23073 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23075 * omp-low.c (struct omp_context): Remove "default_kind" member.
23078 * omp-offload.c (execute_oacc_device_lower): Remove the
23079 parallelism dimensions function attributes for unparallelized
23080 OpenACC kernels constructs.
23082 2017-05-23 Martin Liska <mliska@suse.cz>
23084 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23086 (cgraph_edge::make_speculative): Likewise.
23087 (cgraph_edge::resolve_speculation): Likewise.
23088 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23089 (cgraph_node::dump): Likewise.
23090 * cgraph.h: Likewise.
23091 * cgraphunit.c (analyze_functions): Likewise.
23092 (symbol_table::compile): Likewise.
23093 * ipa-cp.c (print_all_lattices): Likewise.
23094 (determine_versionability): Likewise.
23095 (initialize_node_lattices): Likewise.
23096 (ipcp_verify_propagated_values): Likewise.
23097 (estimate_local_effects): Likewise.
23098 (update_profiling_info): Likewise.
23099 (create_specialized_node): Likewise.
23100 (perhaps_add_new_callers): Likewise.
23101 (decide_about_value): Likewise.
23102 (decide_whether_version_node): Likewise.
23103 (identify_dead_nodes): Likewise.
23104 (ipcp_store_bits_results): Likewise.
23105 * ipa-devirt.c (dump_targets): Likewise.
23106 (ipa_devirt): Likewise.
23107 * ipa-icf.c (sem_item::dump): Likewise.
23108 (sem_function::equals): Likewise.
23109 (sem_variable::equals): Likewise.
23110 (sem_item_optimizer::read_section): Likewise.
23111 (sem_item_optimizer::execute): Likewise.
23112 (congruence_class::dump): Likewise.
23113 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23114 (dump_inline_summary): Likewise.
23115 (estimate_node_size_and_time): Likewise.
23116 (inline_analyze_function): Likewise.
23117 * ipa-inline-transform.c (inline_call): Likewise.
23118 * ipa-inline.c (report_inline_failed_reason): Likewise.
23119 (want_early_inline_function_p): Likewise.
23120 (edge_badness): Likewise.
23121 (update_edge_key): Likewise.
23122 (inline_small_functions): Likewise.
23123 * ipa-profile.c (ipa_profile): Likewise.
23124 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23125 (ipa_make_edge_direct_to_target): Likewise.
23126 (remove_described_reference): Likewise.
23127 (ipa_impossible_devirt_target): Likewise.
23128 (propagate_controlled_uses): Likewise.
23129 (ipa_print_node_params): Likewise.
23130 (ipcp_transform_function): Likewise.
23131 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23132 (propagate_pure_const): Likewise.
23133 * ipa-reference.c (generate_summary): Likewise.
23134 (read_write_all_from_decl): Likewise.
23135 (propagate): Likewise.
23136 (ipa_reference_read_optimization_summary): Likewise.
23137 * ipa-utils.c (ipa_merge_profiles): Likewise.
23138 * ipa.c (walk_polymorphic_call_targets): Likewise.
23139 (symbol_table::remove_unreachable_nodes): Likewise.
23140 (ipa_single_use): Likewise.
23141 * passes.c (execute_todo): Likewise.
23142 * predict.c (drop_profile): Likewise.
23143 * symtab.c (symtab_node::get_dump_name): New function.
23144 (symtab_node::dump_name): Likewise.
23145 (symtab_node::dump_asm_name): Likewise.
23146 (symtab_node::dump_references): Likewise.
23147 (symtab_node::dump_referring): Likewise.
23148 (symtab_node::dump_base): Likewise.
23149 (symtab_node::debug_symtab): Likewise.
23150 * tree-sra.c (convert_callers_for_node): Likewise.
23151 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23152 * value-prof.c (init_node_map): Likewise.
23154 2017-05-23 Martin Liska <mliska@suse.cz>
23156 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23157 and symtab_node::debug_symtab to symbol_table::debug.
23158 * cgraphunit.c (analyze_functions): Use the renamed function.
23159 (symbol_table::compile): Likewise.
23160 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23161 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23162 * passes.c (execute_todo): Likewise.
23163 * symtab.c (symbol_table::dump): New function.
23164 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23166 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23168 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23169 that nonconst implies exec.
23171 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23173 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23174 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23175 (inline_edge_summary_vec): Turn into ...
23176 (ipa_call_summaries): ... this one.
23177 (redirect_to_unreachable, edge_set_predicate,
23178 evaluate_properties_for_edge, inline_summary_alloc,
23179 reset_ipa_call_summary, reset_inline_summary,
23180 inline_summary_t::duplicate): Update.
23181 (inline_edge_duplication_hook): Turn to ...
23182 (ipa_call_summary_t::duplicate): ... this one.
23183 (inline_edge_removal_hook): Turn to ...
23184 (ipa_call_summary_t::remove): ... this one.
23185 (dump_inline_edge_summary): Turn to ...
23186 (dump_ipa_call_summary): ... this one.
23187 (estimate_function_body_sizes): Update.
23188 (inline_update_callee_summaries): Update.
23189 (remap_edge_change_prob): Update.
23190 (remap_edge_summaries): Update.
23191 (inline_merge_summary): Update.
23192 (do_estimate_edge_time): Update.
23193 (inline_generate_summary): Update.
23194 (inline_read_section): Update.
23195 (inline_read_summary): Update.
23196 (inline_free_summary): Update.
23197 * ipa-inline.c (can_inline_edge_p): Update.
23198 (compute_inlined_call_time): Update.
23199 (want_inline_small_function_p): Update.
23200 (edge_badness): Update.
23201 (early_inliner): Update.
23202 * ipa-inline.h (inline_edge_summary): Turn to ...
23203 (ipa_call_summary): ... this one.
23204 (ipa_call_summary_t): New class.
23205 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23206 (ipa_call_summaries): New.
23207 (inline_edge_summary): Remove.
23208 (estimate_edge_growth): Update.
23209 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23210 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23211 * ipa-split.c (execute_split_functions): Update.
23212 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23214 2017-05-23 Tom de Vries <tom@codesourcery.com>
23216 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23217 attributes): Document rdrand effective target.
23219 2017-05-23 Tom de Vries <tom@codesourcery.com>
23221 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23222 attributes): Sort alphabetically.
23224 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23226 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23228 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23231 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23232 V2DF/V2DI splat into two separate patterns, one that handles
23233 registers, and the other that only handles memory. Drop support
23234 for splatting from a GPR on ISA 2.07 and then splitting the
23235 splat into direct move and splat.
23236 (vsx_splat_<mode>_reg): Likewise.
23237 (vsx_splat_<mode>_mem): Likewise.
23239 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23241 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23243 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23245 PR middle-end/80809
23246 * omp-low.c (finish_taskreg_remap): New function.
23247 (finish_taskreg_scan): If unit size of ctx->record_type
23248 is non-constant, unshare the size expression and replace
23249 decls in it with possible outer var refs.
23251 PR middle-end/80809
23252 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23253 GOVD_SHARED rather than GOVD_PRIVATE with it.
23254 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23255 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23257 PR middle-end/80853
23258 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23259 as last argument to build_outer_var_ref for pointer bases of array
23260 section reductions.
23262 2017-05-19 Martin Sebor <msebor@redhat.com>
23264 * print-tree.c (print_node): Print DECL_READ_P flag.
23266 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23268 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23269 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23270 * cgraph.c: Likewise.
23271 * cgraphunit.c: Likewise.
23272 * gengtype.c: Likewise.
23273 * ipa-cp.c: Likewise.
23274 * ipa-devirt.c: Likewise.
23275 * ipa-icf.c: Likewise.
23276 * ipa-predicate.c: Likewise.
23277 * ipa-profile.c: Likewise.
23278 * ipa-prop.c: Likewise.
23279 * ipa-split.c: Likewise.
23281 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23282 edge_predicate_pool, dump_inline_hints,
23283 inline_summary::account_size_time, redirect_to_unreachable,
23284 edge_set_predicate, set_hint_predicate,
23285 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23286 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23287 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23288 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23289 ipa_call_summary_t::remove, initialize_growth_caches,
23290 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23291 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23292 mark_modified, unmodified_parm_1, unmodified_parm,
23293 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23294 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23295 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23296 will_be_nonconstant_predicate, record_modified_bb_info,
23297 get_minimal_bb, record_modified, param_change_prob,
23298 phi_result_unknown_predicate, predicate_for_phi_result,
23299 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23300 estimate_function_body_sizes, compute_inline_parameters,
23301 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23302 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23303 inline_update_callee_summaries, remap_edge_change_prob,
23304 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23305 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23306 inline_analyze_function, inline_summary_t::insert,
23307 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23308 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23309 inline_free_summary): Move to ipa-fnsummary.h
23310 (predicate_t): Remove.
23311 * ipa-fnsummary.c: New file.
23312 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23313 (enum inline_hints_vals, inline_hints, agg_position_info,
23314 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23315 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23316 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23317 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23318 inline_read_summary, inline_write_summary, inline_free_summary,
23319 inline_analyze_function, initialize_inline_failed,
23320 inline_merge_summary, inline_update_overall_summary,
23321 compute_inline_parameters): Move to ipa-fnsummary.h
23322 * ipa-fnsummary.h: New file.
23323 * ipa-inline-transform.h: Include ipa-inline.h.
23324 * ipa-inline.c: LIkewise.
23326 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23328 * ipa-inline.c (edge_badness): Use inlined_time instead of
23329 inline_summaries->get.
23331 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23333 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23335 2017-05-22 Nathan Sidwell <nathan@acm.org>
23337 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23338 (fdump-lang): Document 'raw' option.
23339 * dumpfile.h (TDI_tu): Delete.
23340 * dumpfile.c (dump_files): Remove translation-unit.
23341 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23343 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23345 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23346 command option from $(AWK) call.
23347 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23348 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23349 [FORMAT]: Remove handling of variable.
23350 * config/avr/t-multilib: Regenerate.
23352 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23354 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23356 (dump_inline_summary): Do not print self_time.
23357 (estimate_function_body_sizes): Do not set self_time.
23358 (compute_inline_parameters): Likewise.
23359 (inline_read_section, inline_write_summary): Do not stream self_time.
23360 * ipa-inline.h (inline_summary): Drop self_time.
23362 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23364 * ipa-inline-analysis.c (account_size_time): Rename to ...
23365 (inline_summary::account_size_time): ... this one.
23366 (reset_ipa_call_summary): Turn to ...
23367 (ipa_call_summary::reset): ... this one.
23368 (reset_inline_summary): Turn to ...
23369 (inline_summary::reset): ... this one.
23370 (inline_summary_t::remove): Update.
23371 (inline_summary_t::duplicate): Update.
23372 (ipa_call_summary_t::remove): Update.
23373 (dump_inline_summary): Update.
23374 (estimate_function_body_sizes): Update.
23375 (compute_inline_parameters): Update.
23376 (estimate_node_size_and_time): Update.
23377 (inline_merge_summary): Update.
23378 (inline_update_overall_summary): Update.
23379 (inline_read_section): Update.
23380 (inline_write_summary): Update.
23381 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23382 add account_size_time and reset member functions.
23383 (ipa_call_summary): Add reset function.
23384 * ipa-predicate.h (predicate::operator &): Constify.
23386 2017-05-22 Richard Biener <rguenther@suse.de>
23388 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23390 2017-05-19 Jason Merrill <jason@redhat.com>
23392 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23394 2017-05-19 Marek Polacek <polacek@redhat.com>
23397 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23398 TYPE_OVERFLOW_WRAPS checks.
23400 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23402 * tree-core.h (enum omp_clause_default_kind): Add
23403 "OMP_CLAUSE_DEFAULT_PRESENT".
23404 * tree-pretty-print.c (dump_omp_clause): Handle it.
23405 * gimplify.c (enum gimplify_omp_var_data): Add
23406 "GOVD_MAP_FORCE_PRESENT".
23407 (gimplify_adjust_omp_clauses_1): Map it to
23408 "GOMP_MAP_FORCE_PRESENT".
23409 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23411 * gimplify.c (oacc_default_clause): Clarify.
23413 2017-05-19 Nathan Sidwell <nathan@acm.org>
23415 LANG_HOOK_REGISTER_DUMPS
23416 * toplev.c (general_init): Call register dump lang hook.
23417 * doc/invoke.texi: Document -fdump-lang option family.
23418 * dumpfile.c (dump_files): Remove class dump here.
23419 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23420 * dumpfile.h (tree_dump_index): Remove TDI_class.
23421 * langhooks-def.h (lhd_register_dumps): Declare.
23422 (LANG_HOOKS_REGISTER_DUMPS): Define.
23423 (LANG_HOOKS_INITIALIZER): Add it.
23424 * langhooks.c (lhd_register_dumps): Define.
23425 * langhooks.h (struct lang_hooks): Add register_dumps.
23427 2017-05-19 Nathan Sidwell <nathan@acm.org>
23429 * context.h (context::set_passes): New.
23430 * context.c (context::context): Do not create pass manager.
23431 * toplev.c (general_init): Create pass manager here.
23433 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23435 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23436 use this splitter if two add or or instructions would also work for
23437 the constant we want to generate.
23439 2017-05-19 Richard Biener <rguenther@suse.de>
23442 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23443 predicate evaluation.
23445 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23447 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23449 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23452 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23454 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23455 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23456 (inline_edge_summary_vec): Turn into ...
23457 (ipa_call_summaries): ... this one.
23458 (redirect_to_unreachable, edge_set_predicate,
23459 evaluate_properties_for_edge, inline_summary_alloc,
23460 reset_ipa_call_summary, reset_inline_summary,
23461 inline_summary_t::duplicate): Update.
23462 (inline_edge_duplication_hook): Turn to ...
23463 (ipa_call_summary_t::duplicate): ... this one.
23464 (inline_edge_removal_hook): Turn to ...
23465 (ipa_call_summary_t::remove): ... this one.
23466 (dump_inline_edge_summary): Turn to ...
23467 (dump_ipa_call_summary): ... this one.
23468 (estimate_function_body_sizes): Update.
23469 (inline_update_callee_summaries): Update.
23470 (remap_edge_change_prob): Update.
23471 (remap_edge_summaries): Update.
23472 (inline_merge_summary): Update.
23473 (do_estimate_edge_time): Update.
23474 (inline_generate_summary): Update.
23475 (inline_read_section): Update.
23476 (inline_read_summary): Update.
23477 (inline_free_summary): Update.
23478 * ipa-inline.c (can_inline_edge_p): Update.
23479 (compute_inlined_call_time): Update.
23480 (want_inline_small_function_p): Update.
23481 (edge_badness): Update.
23482 (early_inliner): Update.
23483 * ipa-inline.h (inline_edge_summary): Turn to ...
23484 (ipa_call_summary): ... this one.
23485 (ipa_call_summary_t): New class.
23486 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23487 (ipa_call_summaries): New.
23488 (inline_edge_summary): Remove.
23489 (estimate_edge_growth): Update.
23490 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23491 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23492 * ipa-split.c (execute_split_functions): Update.
23493 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23495 2017-05-19 Richard Biener <rguenther@suse.de>
23497 PR middle-end/80764
23498 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23500 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23502 * config/rs6000/rs6000.c (struct machine_function): Add field
23503 fpr_is_wrapped_separately.
23504 (rs6000_get_separate_components): Use 64 components. Handle the
23505 new FPR components.
23506 (rs6000_components_for_bb): Handle the FPR components.
23507 (rs6000_emit_prologue_components): Handle the FPR components.
23508 (rs6000_emit_epilogue_components): Handle the FPR components.
23509 (rs6000_set_handled_components): Handle the FPR components.
23510 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23511 that are already separately shrink-wrapped.
23512 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23513 that are already separately shrink-wrapped.
23515 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23518 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23521 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23522 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23523 cases where the register allocator uses a move and an offsettable
23524 memory operation to/from a FPR register on ISA 2.06/2.07.
23525 (define_peephole2 for Altivec d-form store): Likewise.
23527 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23530 * config/i386/mmx.md (*mov<mode>_internal): Enable
23531 alternatives 11, 12, 13 and 14 also for 32bit targets.
23532 Remove alternatives 15, 16, 17 and 18.
23533 * config/i386/sse.md (vec_concatv2di): Change
23534 alternative (!x, *y) to (x, ?!*Yn).
23536 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23538 * dumpfile.h (enum dump_kind): Remove stray comma.
23540 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23542 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23543 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23544 predicate::num_conditions
23545 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23546 (CHANGED): turn into predicate::changed.
23547 (agg_position_info): Move to ipa-predicate.h
23548 (add_condition, predicate::add_clause, predicate::operator &=,
23549 predicate::or_with, predicate::evaluate, predicate::probability,
23550 dump_condition, dump_clause, predicate::dump,
23551 predicate::remap_after_duplication, predicate::remap_after_inlining,
23552 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23553 (evaluate_conditions_for_known_args): Update.
23554 (set_cond_stmt_execution_predicate): Update.
23555 * ipa-inline.h: Include ipa-predicate.h
23556 (condition, inline_param_summary, conditions, agg_position_info,
23557 predicate): Move to ipa-predicate.h
23558 * ipa-predicate.c: New file.
23559 * ipa-predicate.h: New file.
23561 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
23563 * final.c (leaf_function_p): Check we are not in a sequence.
23565 2017-05-18 Martin Liska <mliska@suse.cz>
23567 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
23568 * dumpfile.c (dump_register): Use new enum dump_kind.
23569 (get_dump_file_name): Likewise.
23570 (dump_enable_all): Likewise.
23571 (dump_switch_p_1): Likewise.
23572 (enable_rtl_dump_file): Remove usage of TDF_RTL.
23573 * dumpfile.h (enum dump_kind): New enum type.
23574 (struct dump_file_info): Create constructor and
23575 format fields and comments.
23576 * passes.c (pass_manager::register_one_dump_file):
23578 * statistics.c (statistics_early_init): Likewise.
23579 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
23580 TDF_TREE with TDF_SLIM.
23581 (gather_memory_references_ref): Likewise.
23583 2017-05-18 Martin Liska <mliska@suse.cz>
23585 * vec.h (struct vnull): Use it.
23587 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23589 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
23590 (true_predicate, false_predicate, true_predicate_p,
23591 false_predicate_p): Remove.
23592 (single_cond_predicate, not_inlined_predicate): Turn to member function
23594 (add_condition): Update.
23595 (add_clause): Turn to...
23596 (predicate::add_clause): ... this one; update; allow passing NULL
23598 (and_predicates): Turn to ...
23599 (predicate::operator &=): ... this one.
23600 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
23601 (or_predicates): Turn to ...
23602 (predicate::or_with): ... this one.
23603 (evaluate_predicate): Turn to ...
23604 (predicate::evaluate): ... this one.
23605 (predicate_probability): Turn to ...
23606 (predicate::probability): ... this one.
23607 (dump_condition): Update.
23608 (dump_predicate): Turn to ...
23609 (predicate::dump): ... this one.
23610 (account_size_time): Update.
23611 (edge_set_predicate): Update.
23612 (set_hint_predicate): UPdate.
23613 (evaluate_conditions_for_known_args): Update.
23614 (evaluate_properties_for_edge): Update.
23615 (remap_predicate_after_duplication): Turn to...
23616 (predicate::remap_after_duplication): ... this one.
23617 (remap_hint_predicate_after_duplication): Update.
23618 (inline_summary_t::duplicate): UPdate.
23619 (dump_inline_edge_summary): Update.
23620 (dump_inline_summary): Update.
23621 (set_cond_stmt_execution_predicate): Update.
23622 (set_switch_stmt_execution_predicate): Update.
23623 (compute_bb_predicates): Update.
23624 (will_be_nonconstant_expr_predicate): Update.
23625 (will_be_nonconstant_predicate): Update.
23626 (phi_result_unknown_predicate): Update.
23627 (predicate_for_phi_result): Update.
23628 (array_index_predicate): Update.
23629 (estimate_function_body_sizes): Update.
23630 (estimate_node_size_and_time): Update.
23631 (estimate_ipcp_clone_size_and_time): Update.
23632 (remap_predicate): Rename to ...
23633 (predicate::remap_after_inlining): ... this one.
23634 (remap_hint_predicate): Update.
23635 (inline_merge_summary): Update.
23636 (inline_update_overall_summary): Update.
23637 (estimate_size_after_inlining): Update.
23638 (read_predicate): Rename to ...
23639 (predicate::stream_in): ... this one.
23640 (read_inline_edge_summary): Update.
23641 (write_predicate): Rename to ...
23642 (predicate::stream_out): ... this one.
23643 (write_inline_edge_summary): Update.
23644 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
23645 (clause_t): Turn to uint32_t
23646 (predicate): Turn to class; implement constructor and operators
23648 (size_time_entry): Update.
23649 (inline_summary): Update.
23650 (inline_edge_summary): Update.
23652 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
23654 * fold-const.c (fold_binary_loc): Move transformation...
23655 * match.pd (C - X CMP X): ... here.
23657 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
23659 * config/sparc/sparc.c (sparc_option_override): Set function
23660 alignment for -mcpu=niagara7 to 64 to match the I$ line.
23661 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
23663 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
23665 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
23667 2017-05-18 Marek Polacek <polacek@redhat.com>
23670 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
23671 (pass_ubsan::execute): Call gimple_assign_single_p instead of
23672 gimple_assign_load_p.
23674 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
23676 PR middle-end/80692
23677 * real.c (do_compare): Give decimal_do_compare preference over
23678 comparing just the signs.
23680 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
23682 * doc/md.texi (Canonicalization of Instructions): Describe the
23683 canonical form of instructions that inherently set a condition
23686 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
23688 PR middle-end/80775
23689 * tree-cfg.c: Move deletion of unreachable case statements to after
23690 the merging of consecutive case labels.
23692 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
23694 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
23695 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
23696 restoring of callee-saved registers.
23698 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
23700 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
23701 * config/visium/visium.c (single_set_and_flags): Likewise.
23702 * config/visium/visium.md (Substitutions): Likewise.
23704 2017-05-17 Martin Liska <mliska@suse.cz>
23706 * cfg.c: Introduce dump_flags_t type and
23707 use it instead of int type.
23709 * cfghooks.c: Likewise.
23710 * cfghooks.h (struct cfg_hooks): Likewise.
23711 * cfgrtl.c: Likewise.
23712 * cfgrtl.h: Likewise.
23713 * cgraph.c (cgraph_node::get_body): Likewise.
23714 * coretypes.h: Likewise.
23715 * domwalk.c: Likewise.
23716 * domwalk.h: Likewise.
23717 * dumpfile.c (struct dump_option_value_info): Likewise.
23718 (dump_enable_all): Likewise.
23719 (dump_switch_p_1): Likewise.
23720 (opt_info_switch_p): Likewise.
23721 * dumpfile.h (enum tree_dump_index): Likewise.
23722 (struct dump_file_info): Likewise.
23723 * genemit.c: Likewise.
23724 * generic-match-head.c: Likewise.
23725 * gengtype.c (open_base_files): Likewise.
23726 * gimple-pretty-print.c: Likewise.
23727 * gimple-pretty-print.h: Likewise.
23728 * graph.c (print_graph_cfg): Likewise.
23729 * graphite-scop-detection.c (dot_all_sese): Likewise.
23730 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
23731 * loop-unroll.c (report_unroll): Likewise.
23732 * passes.c (pass_manager::register_one_dump_file): Likewise.
23733 * print-tree.c: Likewise.
23734 * statistics.c: Likewise.
23735 * tree-cfg.c: Likewise.
23736 * tree-cfg.h: Likewise.
23737 * tree-dfa.c: Likewise.
23738 * tree-dfa.h: Likewise.
23739 * tree-dump.c (dump_function): Likewise.
23740 * tree-dump.h (struct dump_info): Likewise.
23741 * tree-pretty-print.c: Likewise.
23742 * tree-pretty-print.h: Likewise.
23743 * tree-ssa-live.c: Likewise.
23744 * tree-ssa-live.h: Likewise.
23745 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23746 * tree-vect-loop.c: Likewise.
23747 * tree-vect-slp.c: Likewise.
23749 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23750 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23752 PR tree-optimization/80457
23753 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23754 of all arguments to a statement as scalar_to_vec operations.
23755 (vectorizable_call): Adjust call to vect_model_simple_cost for
23757 (vectorizable_conversion): Likewise.
23758 (vectorizable_assignment): Likewise.
23759 (vectorizable_shift): Likewise.
23760 (vectorizable_operation): Likewise.
23761 (vectorizable_comparison): Likewise.
23762 (vect_is_simple_cond): Record the def types for operands.
23763 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23764 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23765 for statement argument count.
23767 2017-05-16 Carl Love <cel@us.ibm.com>
23769 * config/rs6000/rs6000-c: Add support for built-in functions
23770 vector unsigned long long vec_bperm (vector unsigned long long,
23771 vector unsigned char)
23772 vector signed long long vec_mule (vector signed int,
23774 vector unsigned long long vec_mule (vector unsigned int,
23775 vector unsigned int)
23776 vector signed long long vec_mulo (vector signed int,
23778 vector unsigned long long vec_mulo (vector unsigned int,
23779 vector unsigned int)
23780 vector signed char vec_sldw (vector signed char,
23781 vector signed char,
23783 vector unsigned char vec_sldw (vector unsigned char,
23784 vector unsigned char,
23786 vector signed short vec_sldw (vector signed short,
23787 vector signed short,
23789 vector unsigned short vec_sldw (vector unsigned short,
23790 vector unsigned short,
23792 vector signed int vec_sldw (vector signed int,
23795 vector unsigned int vec_sldw (vector unsigned int,
23796 vector unsigned int,
23798 vector signed long long vec_sldw (vector signed long long,
23799 vector signed long long,
23801 vector unsigned long long vec_sldw (vector unsigned long long,
23802 vector unsigned long long,
23804 * config/rs6000/rs6000-c: Add support for built-in functions
23805 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
23806 * config/rs6000/altivec.h: Add defintion for vec_sldw.
23807 * doc/extend.texi: Update the built-in documentation for the
23808 new built-in functions.
23810 2017-05-16 Marek Polacek <polacek@redhat.com>
23814 * tree.c (save_expr): Don't fold the expression.
23816 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
23818 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23819 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23820 and (?*y,m). Update insn attributes.
23822 2017-05-16 Martin Liska <mliska@suse.cz>
23824 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23825 flags argument of print_gimple_stmt, print_gimple_expr,
23826 print_generic_stmt and print_generic_expr.
23827 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23828 * coretypes.h: Likewise.
23829 * except.c (dump_eh_tree): Likewise.
23830 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23831 * gimple-pretty-print.h: Likewise.
23832 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23833 (backprop::push_to_worklist): Likewise.
23834 (backprop::pop_from_worklist): Likewise.
23835 (backprop::process_use): Likewise.
23836 (backprop::intersect_uses): Likewise.
23837 (note_replacement): Likewise.
23838 * gimple-ssa-store-merging.c
23839 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23840 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23841 (pass_store_merging::execute): Likewise.
23842 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23843 (ssa_base_cand_dump_callback): Likewise.
23844 (dump_incr_vec): Likewise.
23845 (replace_refs): Likewise.
23846 (replace_mult_candidate): Likewise.
23847 (create_add_on_incoming_edge): Likewise.
23848 (create_phi_basis): Likewise.
23849 (insert_initializers): Likewise.
23850 (all_phi_incrs_profitable): Likewise.
23851 (introduce_cast_before_cand): Likewise.
23852 (replace_one_candidate): Likewise.
23853 * gimplify.c (gimplify_expr): Likewise.
23854 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23855 (set_rename): Likewise.
23856 (rename_uses): Likewise.
23857 (copy_loop_phi_nodes): Likewise.
23858 (add_close_phis_to_merge_points): Likewise.
23859 (copy_loop_close_phi_args): Likewise.
23860 (copy_cond_phi_args): Likewise.
23861 (graphite_copy_stmts_from_block): Likewise.
23862 (translate_pending_phi_nodes): Likewise.
23863 * graphite-poly.c (print_pdr): Likewise.
23864 (dump_gbb_cases): Likewise.
23865 (dump_gbb_conditions): Likewise.
23866 (print_scop_params): Likewise.
23867 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23868 (build_cross_bb_scalars_use): Likewise.
23869 (gather_bbs::before_dom_children): Likewise.
23870 * hsa-dump.c (dump_hsa_immed): Likewise.
23871 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23872 (get_replacement_map): Likewise.
23873 * ipa-inline-analysis.c (dump_condition): Likewise.
23874 (estimate_function_body_sizes): Likewise.
23875 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23876 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23877 * ipa-prop.c (ipa_dump_param): Likewise.
23878 (ipa_print_node_jump_functions_for_edge): Likewise.
23879 (ipa_modify_call_arguments): Likewise.
23880 (ipa_modify_expr): Likewise.
23881 (ipa_dump_param_adjustments): Likewise.
23882 (ipa_dump_agg_replacement_values): Likewise.
23883 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23884 * ipa-pure-const.c (check_stmt): Likewise.
23885 (pass_nothrow::execute): Likewise.
23886 * ipa-split.c (execute_split_functions): Likewise.
23887 * omp-offload.c (dump_oacc_loop_part): Likewise.
23888 (dump_oacc_loop): Likewise.
23889 * trans-mem.c (tm_log_emit): Likewise.
23890 (tm_memopt_accumulate_memops): Likewise.
23891 (dump_tm_memopt_set): Likewise.
23892 (dump_tm_memopt_transform): Likewise.
23893 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23894 (print_loop): Likewise.
23895 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23896 (chkp_gather_checks_info): Likewise.
23897 (chkp_get_check_result): Likewise.
23898 (chkp_remove_check_if_pass): Likewise.
23899 (chkp_use_outer_bounds_if_possible): Likewise.
23900 (chkp_reduce_bounds_lifetime): Likewise.
23901 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23902 (chkp_mark_completed_bounds): Likewise.
23903 (chkp_register_incomplete_bounds): Likewise.
23904 (chkp_mark_invalid_bounds): Likewise.
23905 (chkp_maybe_copy_and_register_bounds): Likewise.
23906 (chkp_build_returned_bound): Likewise.
23907 (chkp_get_bound_for_parm): Likewise.
23908 (chkp_build_bndldx): Likewise.
23909 (chkp_get_bounds_by_definition): Likewise.
23910 (chkp_generate_extern_var_bounds): Likewise.
23911 (chkp_get_bounds_for_decl_addr): Likewise.
23912 * tree-chrec.c (chrec_apply): Likewise.
23913 * tree-data-ref.c (dump_data_reference): Likewise.
23914 (dump_subscript): Likewise.
23915 (dump_data_dependence_relation): Likewise.
23916 (analyze_overlapping_iterations): Likewise.
23917 * tree-inline.c (expand_call_inline): Likewise.
23918 (tree_function_versioning): Likewise.
23919 * tree-into-ssa.c (dump_defs_stack): Likewise.
23920 (dump_currdefs): Likewise.
23921 (dump_names_replaced_by): Likewise.
23922 (dump_update_ssa): Likewise.
23923 (update_ssa): Likewise.
23924 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23925 * tree-parloops.c (build_new_reduction): Likewise.
23926 (try_create_reduction_list): Likewise.
23927 (ref_conflicts_with_region): Likewise.
23928 (oacc_entry_exit_ok_1): Likewise.
23929 (oacc_entry_exit_single_gang): Likewise.
23930 * tree-pretty-print.h: Likewise.
23931 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23932 (get_scalar_evolution): Likewise.
23933 (add_to_evolution): Likewise.
23934 (get_loop_exit_condition): Likewise.
23935 (analyze_evolution_in_loop): Likewise.
23936 (analyze_initial_condition): Likewise.
23937 (analyze_scalar_evolution): Likewise.
23938 (instantiate_scev): Likewise.
23939 (number_of_latch_executions): Likewise.
23940 (gather_chrec_stats): Likewise.
23941 (final_value_replacement_loop): Likewise.
23942 (scev_const_prop): Likewise.
23943 * tree-sra.c (dump_access): Likewise.
23944 (disqualify_candidate): Likewise.
23945 (create_access): Likewise.
23946 (reject): Likewise.
23947 (maybe_add_sra_candidate): Likewise.
23948 (create_access_replacement): Likewise.
23949 (analyze_access_subtree): Likewise.
23950 (analyze_all_variable_accesses): Likewise.
23951 (sra_modify_assign): Likewise.
23952 (initialize_constant_pool_replacements): Likewise.
23953 (find_param_candidates): Likewise.
23954 (decide_one_param_reduction): Likewise.
23955 (replace_removed_params_ssa_names): Likewise.
23956 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23957 * tree-ssa-copy.c (dump_copy_of): Likewise.
23958 (copy_prop_visit_cond_stmt): Likewise.
23959 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23960 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23961 (record_equivalences_from_stmt): Likewise.
23962 * tree-ssa-dse.c (compute_trims): Likewise.
23963 (delete_dead_call): Likewise.
23964 (delete_dead_assignment): Likewise.
23965 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23966 (forward_propagate_into_cond): Likewise.
23967 (pass_forwprop::execute): Likewise.
23968 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23969 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23971 (move_computations_worker): Likewise.
23972 (execute_sm): Likewise.
23973 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23974 (remove_exits_and_undefined_stmts): Likewise.
23975 (remove_redundant_iv_tests): Likewise.
23976 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23977 (adjust_iv_update_pos): Likewise.
23978 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23979 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23980 (value_replacement): Likewise.
23981 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23982 * tree-ssa-pre.c (print_pre_expr): Likewise.
23983 (get_representative_for): Likewise.
23984 (create_expression_by_pieces): Likewise.
23985 (insert_into_preds_of_block): Likewise.
23986 (eliminate_insert): Likewise.
23987 (eliminate_dom_walker::before_dom_children): Likewise.
23988 (eliminate): Likewise.
23989 (remove_dead_inserted_code): Likewise.
23990 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23991 * tree-ssa-reassoc.c (get_rank): Likewise.
23992 (eliminate_duplicate_pair): Likewise.
23993 (eliminate_plus_minus_pair): Likewise.
23994 (eliminate_not_pairs): Likewise.
23995 (undistribute_ops_list): Likewise.
23996 (eliminate_redundant_comparison): Likewise.
23997 (update_range_test): Likewise.
23998 (optimize_range_tests_var_bound): Likewise.
23999 (optimize_vec_cond_expr): Likewise.
24000 (rewrite_expr_tree): Likewise.
24001 (rewrite_expr_tree_parallel): Likewise.
24002 (linearize_expr): Likewise.
24003 (break_up_subtract): Likewise.
24004 (linearize_expr_tree): Likewise.
24005 (attempt_builtin_powi): Likewise.
24006 (attempt_builtin_copysign): Likewise.
24007 (transform_stmt_to_copy): Likewise.
24008 (transform_stmt_to_multiply): Likewise.
24009 (dump_ops_vector): Likewise.
24010 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24011 (print_scc): Likewise.
24012 (set_ssa_val_to): Likewise.
24013 (visit_reference_op_store): Likewise.
24014 (visit_use): Likewise.
24015 (sccvn_dom_walker::before_dom_children): Likewise.
24016 (run_scc_vn): Likewise.
24017 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24019 (expr_hash_elt::print): Likewise.
24020 (const_and_copies::pop_to_marker): Likewise.
24021 (const_and_copies::record_const_or_copy_raw): Likewise.
24022 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24023 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24024 (dump_predicates): Likewise.
24025 (find_uninit_use): Likewise.
24026 (warn_uninitialized_phi): Likewise.
24027 (pass_late_warn_uninitialized::execute): Likewise.
24028 * tree-ssa.c (verify_vssa): Likewise.
24029 (verify_ssa): Likewise.
24030 (maybe_optimize_var): Likewise.
24031 * tree-vrp.c (dump_value_range): Likewise.
24032 (dump_all_value_ranges): Likewise.
24033 (dump_asserts_for): Likewise.
24034 (register_edge_assert_for_2): Likewise.
24035 (vrp_visit_cond_stmt): Likewise.
24036 (vrp_visit_switch_stmt): Likewise.
24037 (vrp_visit_stmt): Likewise.
24038 (vrp_visit_phi_node): Likewise.
24039 (simplify_cond_using_ranges_1): Likewise.
24040 (fold_predicate_in): Likewise.
24041 (evrp_dom_walker::before_dom_children): Likewise.
24042 (evrp_dom_walker::push_value_range): Likewise.
24043 (evrp_dom_walker::pop_value_range): Likewise.
24044 (execute_early_vrp): Likewise.
24046 2017-05-16 Richard Biener <rguenther@suse.de>
24048 * dwarf2out.c (loc_list_from_tree_1): Do not create
24049 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24051 2017-05-16 Richard Biener <rguenther@suse.de>
24053 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24055 (note_variable_value_in_expr): If we resolved the decl ref
24056 do not push to the stack.
24058 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24060 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24061 operations in fast-math mode.
24062 (vaddq_f16): Likewise.
24063 (vmul_f16): Likewise.
24064 (vmulq_f16): Likewise.
24065 (vsub_f16): Likewise.
24066 (vsubq_f16): Likewise.
24067 * config/arm/neon.md (add<mode>3): New.
24069 (fma:<VH:mode>3): New. Also remove outdated comment.
24072 2017-05-16 Martin Liska <mliska@suse.cz>
24076 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24077 (odr_types_equivalent_p): Likewise.
24079 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24081 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24083 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24086 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24087 non-interunit SSE move alternatives with '?'.
24088 (zero-extendsidi peephole2): New peephole to skip intermediate
24089 general register in SSE zero-extend sequence.
24091 2017-05-15 Jeff Law <law@redhat.com>
24093 * reorg.c (relax_delay_slots): Create a new variable to hold
24094 the temporary target rather than clobbering TARGET_LABEL.
24096 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24097 missing argument to extract_bit_field call.
24098 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24100 2017-05-15 Martin Liska <mliska@suse.cz>
24103 * gcc.c (process_command): Do not allow empty argument of -o option.
24105 2017-05-15 Renlin Li <renlin.li@arm.com>
24107 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24108 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24109 * config/aarch64/constraints.md (Usf): Add long call check.
24110 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24111 (call_value): Likewise.
24112 (sibcall): Likewise.
24113 (sibcall_value): Likewise.
24115 (call_value_insn): New.
24116 (sibcall_insn): Update rtx pattern.
24117 (sibcall_value_insn): Likewise.
24118 (call_internal): Remove.
24119 (call_value_internal): Likewise.
24120 (sibcall_internal): Likewise.
24121 (sibcall_value_internal): Likewise.
24122 (call_reg): Likewise.
24123 (call_symbol): Likewise.
24124 (call_value_reg): Likewise.
24125 (call_value_symbol): Likewise.
24127 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24130 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24132 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24134 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24135 compatible with CCGOCmode and with CCZmode.
24137 2017-05-14 Martin Sebor <msebor@redhat.com>
24139 PR middle-end/77671
24140 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24141 (gimple_fold_builtin_snprintf): Same.
24142 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24143 (gimple_fold_builtin_snprintf): Same.
24144 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24145 of character types.
24146 (is_call_safe): New function.
24147 (try_substitute_return_value): Call it.
24148 (try_simplify_call): New function.
24149 (pass_sprintf_length::handle_gimple_call): Call it.
24151 2017-05-14 Martin Sebor <msebor@redhat.com>
24153 PR middle-end/80669
24154 * builtins.c (expand_builtin_stpncpy): Simplify.
24156 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24158 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24159 * config/i386/i386.h
24160 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24161 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24162 (struct machine_function): Add new members call_ms2sysv,
24163 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24164 (struct machine_frame_state): New fields sp_realigned and
24165 sp_realigned_offset.
24166 * config/i386/i386.c
24167 (enum xlogue_stub): New enum.
24168 (enum xlogue_stub_sets): New enum.
24169 (class xlogue_layout): New class.
24170 (struct ix86_frame): New fields stack_realign_allocate_offset,
24171 stack_realign_offset and outlined_save_offset. Modify comments to
24172 detail stack layout when using out-of-line stubs.
24173 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24174 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24175 -mcall-ms2sysv-xlogues.
24176 (stub_managed_regs): New static variable.
24177 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24178 registers managed by out-of-line stub.
24179 (disable_call_ms2sysv_xlogues): New function.
24180 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24181 m->call_ms2sysv when appropriate and compute frame layout for
24183 (sp_valid_at, fp_valid_at): New inline functions.
24184 (choose_basereg): New function.
24185 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24187 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24188 Use align parameter of choose_baseaddr to generated aligned SSE movs
24190 (pro_epilogue_adjust_stack): Modify to track
24191 machine_frame_state::sp_realigned.
24192 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24193 (ix86_nsaved_sseregs): Likewise.
24194 (ix86_emit_save_regs): Likewise.
24195 (ix86_emit_save_regs_using_mov): Likewise.
24196 (ix86_emit_save_sse_regs_using_mov): Likewise.
24197 (get_scratch_register_on_entry): Likewise.
24198 (gen_frame_set): New function.
24199 (gen_frame_load): Likewise.
24200 (gen_frame_store): Likewise.
24201 (emit_outlined_ms2sysv_save): Likewise.
24202 (emit_outlined_ms2sysv_restore): Likewise.
24203 (ix86_expand_prologue): Modify stack re-alignment code and call
24204 emit_outlined_ms2sysv_save when appropriate.
24205 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24206 parameter rtx_insn *insn, which allows the function to be used to only
24207 generate the notes.
24208 (ix86_expand_epilogue): Modify validity checks of frame and stack
24209 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24210 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24211 * config/i386/predicates.md
24212 (save_multiple): New predicate.
24213 (restore_multiple): Likewise.
24214 * config/i386/sse.md
24215 (save_multiple<mode>): New pattern.
24216 (save_multiple_realign<mode>): Likewise.
24217 (restore_multiple<mode>): Likewise.
24218 (restore_multiple_and_return<mode>): Likewise.
24219 (restore_multiple_leave_return<mode>): Likewise.
24220 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24222 2017-05-14 Julia Koval <julia.koval@intel.com>
24224 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24225 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24226 (__builtin_ia32_xsetbv): New builtins.
24227 * config/i386/i386.c (ix86_expand_special_args_builtin):
24229 (ix86_expand_builtin): Special expand for new intrinsics.
24230 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24231 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24232 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24234 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24236 * cfganal.c (inverted_post_order_compute): Change argument type
24238 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24239 * df-core.c (rest_of_handle_df_initialize): Adjust.
24240 (rest_of_handle_df_finish): Likewise.
24241 (df_analyze_1): Likewise.
24242 (df_analyze): Likewise.
24243 (loop_inverted_post_order_compute): Change argument to be a vec *.
24244 (df_analyze_loop): Adjust.
24245 (df_get_n_blocks): Likewise.
24246 (df_get_postorder): Likewise.
24247 * df.h (struct df_d): Change field to be a vec.
24248 * lcm.c (compute_laterin): Adjust.
24249 (compute_available): Likewise.
24250 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24251 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24252 * tree-ssa-pre.c (compute_antic): Likewise.
24254 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24256 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24257 (depth_first_search::depth_first_search): Change structure init
24258 function to this constructor.
24259 (depth_first_search::add_bb): Rename function to this member.
24260 (depth_first_search::execute): Likewise.
24261 (flow_dfs_compute_reverse_finish): Adjust.
24263 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24265 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24266 (longest_simple_path): Likewise.
24267 * shrink-wrap.c (spread_components): Likewise.
24268 (disqualify_problematic_components): Likewise.
24269 (emit_common_heads_for_components): Likewise.
24270 (emit_common_tails_for_components): Likewise.
24271 (insert_prologue_epilogue_for_components): Likewise.
24273 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24275 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24278 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24280 * df-core.c (df_set_blocks): Start using auto_bitmap.
24281 (df_compact_blocks): Likewise.
24282 * df-problems.c (df_rd_confluence_n): Likewise.
24283 * df-scan.c (df_insn_rescan_all): Likewise.
24284 (df_process_deferred_rescans): Likewise.
24285 (df_update_entry_block_defs): Likewise.
24286 (df_update_exit_block_uses): Likewise.
24287 (df_entry_block_bitmap_verify): Likewise.
24288 (df_exit_block_bitmap_verify): Likewise.
24289 (df_scan_verify): Likewise.
24290 * lra-constraints.c (lra_constraints): Likewise.
24291 (undo_optional_reloads): Likewise.
24292 (lra_undo_inheritance): Likewise.
24293 * lra-remat.c (calculate_gen_cands): Likewise.
24294 (do_remat): Likewise.
24295 * lra-spills.c (assign_spill_hard_regs): Likewise.
24296 (spill_pseudos): Likewise.
24297 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24298 (bitmap_set_subtract_values): Likewise.
24300 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24302 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24303 management with auto_bitmap.
24304 (fix_inter_tick): Likewise.
24305 (fix_recovery_deps): Likewise.
24306 * ira.c (add_store_equivs): Likewise.
24307 (find_moveable_pseudos): Likewise.
24308 (split_live_ranges_for_shrink_wrap): Likewise.
24309 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24310 (rtx_reuse_manager::seen_def_p): Likewise.
24311 (rtx_reuse_manager::set_seen_def): Likewise.
24312 * print-rtl.h (class rtx_reuse_manager): Likewise.
24314 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24316 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24318 (migrate_btr_def): Likewise.
24319 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24320 * df-core.c (loop_post_order_compute): Likewise.
24321 (loop_inverted_post_order_compute): Likewise.
24322 * hsa-common.h: Likewise.
24323 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24324 * init-regs.c (initialize_uninitialized_regs): Likewise.
24325 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24326 (inline_small_functions): Likewise.
24327 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24328 * ira.c (combine_and_move_insns): Likewise.
24329 (build_insn_chain): Likewise.
24330 * loop-invariant.c (find_invariants): Likewise.
24331 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24332 * predict.c (tree_predict_by_opcode): Likewise.
24333 (predict_paths_leading_to): Likewise.
24334 (predict_paths_leading_to_edge): Likewise.
24335 (estimate_loops_at_level): Likewise.
24336 (estimate_loops): Likewise.
24337 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24338 (spread_components): Likewise.
24339 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24340 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24341 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24342 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24343 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24344 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24345 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24346 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24347 (mark_threaded_blocks): Likewise.
24348 (thread_through_all_blocks): Likewise.
24349 * tree-ssa.c (verify_ssa): Likewise.
24350 (execute_update_addresses_taken): Likewise.
24351 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24353 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24355 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24357 (post_order_compute): Likewise.
24358 (inverted_post_order_compute): Likewise.
24359 (pre_and_rev_post_order_compute_fn): Likewise.
24361 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24363 * genrecog.c (int_set::int_set): Explicitly construct our
24364 auto_vec base class.
24365 * vec.h (auto_vec::auto_vec): New constructor.
24367 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24369 * bitmap.h (class auto_bitmap): New constructor taking
24370 bitmap_obstack * argument.
24372 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24374 * bitmap.h (class auto_bitmap): Change type of m_bits to
24375 bitmap_head, and adjust ctor / dtor and member operators.
24377 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24379 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24380 when returned register mode doesn't match original mode.
24382 2017-05-12 Jeff Law <law@redhat.com>
24383 Jakub Jelinek <jakub@redhat.com>
24385 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24386 we look for cc setter after the compare-elim changes.
24387 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24388 within the vector to match what compare-elim now expects.
24389 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24390 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24392 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24393 after the compare-elim changes.
24394 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24395 the vector to match what compare-elim now expects.
24396 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24397 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24398 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24399 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24400 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24402 * config/visium/visium.c (single_set_and_flags): Fix where
24403 we look for cc setter after the compare-elim changes.
24404 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24405 with the vector to match what compare-elim now expects.
24406 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24407 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24408 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24409 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24410 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24411 (neg<mode>2_insn_set_overflow): Likewise.
24413 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24415 PR middle-end/79794
24416 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24417 maybe_expand_insn call, set ops[0].target. If still set after call,
24418 set alt_rtl. Add extra arg to recursive calls.
24419 (extract_bit_field): Add alt_rtl argument. Pass to
24421 * expmed.h (extract_bit_field): Fix prototype.
24422 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24423 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24424 to extract_bit_field_calls.
24425 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24426 Pass alt_rtl to extract_bit_field calls.
24427 * calls.c (store_unaligned_arguments_into_psuedos)
24428 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24429 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24431 * optabs.h (struct expand_operand): Add target bitfield.
24433 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24435 * compare-elim.c (try_eliminate_compare): Canonicalize
24436 operation with embedded compare to
24437 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24438 (set (reg) (operation)].
24440 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24442 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24445 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24446 cost of adding a carry flag for ADC instruction.
24447 [case MINUS]: Ignore the cost of subtracting a carry flag
24448 for SBB instruction.
24450 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24452 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24454 * config/rs6000/bmiintrin.h: New file.
24455 * config/rs6000/bmi2intrin.h: New file.
24456 * config/rs6000/x86intrin.h: New file.
24458 2017-05-12 Jeff Law <law@redhat.com>
24460 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24463 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24465 PR middle-end/80707
24466 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24468 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24470 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24471 early expansion of vector divide builtins.
24472 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24473 builtins identified as having unsigned arguments.
24475 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24477 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24478 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24479 expansion of vector logical operations (and, andc, or, xor,
24482 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24484 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24485 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24487 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24489 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24490 early GIMPLE expansion of vector multiplies.
24492 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24494 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24495 TARGET_HAVE_MOVT conditional.
24496 (movt splitter): Likewise.
24498 2017-05-12 Richard Biener <rguenther@suse.de>
24500 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24501 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24502 Fold all stmts not inplace.
24504 2017-05-12 Richard Biener <rguenther@suse.de>
24506 PR tree-optimization/80713
24507 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24508 inserted_exprs bit for not removed stmts.
24510 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24512 PR middle-end/69921
24513 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24514 parallelized" attribute for parallelized OpenACC kernels.
24515 * omp-offload.c (execute_oacc_device_lower): Use it.
24517 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24518 Set "oacc kernels" attribute.
24519 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24520 parameter. Adjust all users.
24521 (oacc_fn_attrib_kernels_p): Remove function.
24522 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24523 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24524 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24525 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24526 assert "oacc kernels" attribute is set.
24528 2017-05-11 Carl Love <cel@us.ibm.com>
24530 * config/rs6000/rs6000-c: Add support for built-in functions
24531 vector unsigned char vec_popcnt (vector signed char)
24532 vector unsigned char vec_popcnt (vector unsigned char)
24533 vector unsigned short vec_popcnt (vector signed short)
24534 vector unsigned short vec_popcnt (vector unsigned short)
24535 vector unsigned int vec_popcnt (vector signed int)
24536 vector unsigned int vec_popcnt (vector unsigned int)
24537 vector unsigned long long vec_popcnt (vector signed long long)
24538 vector unsigned long long vec_popcnt (vector unsigned long long)
24539 vector signed long long vec_slo (vector signed long long,
24540 vector signed char)
24541 vector signed long long vec_slo (vector signed long long,
24542 vector unsigned char)
24543 vector unsigned long long vec_slo (vector unsigned long long,
24544 vector signed char)
24545 vector unsigned long long vec_slo (vector unsigned long long,
24546 vector unsigned char)
24547 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24548 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24549 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24550 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24551 * doc/extend.texi: Update the built-in documentation file for the
24552 new built-in functions.
24554 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
24556 * attribs.h (sorted_attr_string): Move machine independent
24557 functions for target clone support from the i386 port to common
24558 code. Rename ix86_function_versions to common_function_versions.
24559 Rename make_name to make_unique_name.
24560 (common_function_versions): Likewise.
24561 (make_unique_name): Likewise.
24562 (make_dispatcher_decl): Likewise.
24563 (is_function_default_version): Likewise.
24564 * attribs.c (attr_strcmp): Likewise.
24565 (sorted_attr_string): Likewise.
24566 (common_function_versions): Likewise.
24567 (make_unique_name): Likewise.
24568 (make_dispatcher_decl): Likewise.
24569 (is_function_default_version): Likewise.
24570 * config/i386/i386.c (attr_strcmp): Likewise.
24571 (sorted_attr_string): Likewise.
24572 (ix86_function_versions): Likewise.
24573 (make_name): Likewise.
24574 (make_dispatcher_decl): Likewise.
24575 (is_function_default_version): Likewise.
24576 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
24578 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24581 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
24582 Account for direct move costs for vec_construct of integer
24585 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
24588 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
24589 (UNSPEC_STX_ATOMIC): Ditto.
24590 (loaddi_via_sse): New insn.
24591 (storedi_via_sse): Ditto.
24592 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
24593 Update corresponding peephole2 patterns.
24594 (atomic_storedi_fpu): Ditto.
24596 2017-05-11 Julia Koval <julia.koval@intel.com>
24598 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
24599 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
24601 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
24602 (__builtin_ia32_rsqrt14ss_mask): New builtins.
24603 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
24605 2017-05-11 Nathan Sidwell <nathan@acm.org>
24607 * graphite-poly.c: Include dumpfile.h.
24609 * dumpfle.h (dump_function): Declare here ...
24610 * tree-dump.h (dump_function): ... not here.
24611 * dumpfile.c: #include tree-cfg.h.
24612 (dump_function): Move here from ...
24613 * tree-dump.c (dump_function): ... here.
24614 * gimplify.c: #include splay-tree.h, not tree-dump.h.
24615 * graphite-poly.c: Don't include tree-dump.h.
24616 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
24617 * print-tree.c: Likewise.
24618 * stor-layout.c: Likewise.
24619 * tree-nested.c: Likewise.
24621 * dumpfile.c (dump_start): Use TDF_FLAGS.
24622 (dump_enable_all): Fix TDF_KIND check thinko.
24624 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
24626 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24627 array entries to represent two legal parameterizations of the
24628 overloaded __builtin_cmpb function, as represented by the
24629 P6_OV_BUILTIN_CMPB constant.
24630 (altivec_resolve_overloaded_builtin): Add special case handling
24631 for the __builtin_cmpb function, as represented by the
24632 P6_OV_BUILTIN_CMPB constant.
24633 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
24634 (BU_P6_64BIT_2): New macro.
24635 (BU_P6_OVERLOAD_2): New macro
24636 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
24637 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
24638 (CMPB): Add overload support to represent both 32-bit and 64-bit
24639 compare-bytes function.
24640 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
24641 support for TARGET_CMPB.
24642 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
24643 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24644 documentation of the __builtin_cmpb overloaded built-in function.
24646 2017-05-11 Richard Biener <rguenther@suse.de>
24648 PR tree-optimization/80705
24649 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
24650 bases are not vectorizable.
24652 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24654 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
24655 when counting register pressure.
24657 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24659 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
24660 (iv_ca_more_deps): Renamed to ...
24661 (iv_ca_compare_deps): ... this.
24662 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
24664 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24666 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
24668 (determine_group_iv_costs): ... here.
24669 (find_inv_vars_cb): Record inv var if it's not recorded before.
24671 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24673 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
24674 (get_shiftadd_cost): Ditto.
24676 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24678 * tree-ssa-address.c: Include header file.
24679 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
24681 (add_to_parts): Refactor.
24682 (addr_to_parts): New parameter. Update use of move_hint_to_base.
24683 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
24686 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24688 PR tree-optimization/53090
24689 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
24691 (extract_cond_operands): Detect condition with IV on both sides
24692 and return COMP_IV_EXPR_2.
24693 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
24694 (rewrite_use_compare): Simplify by removing call to function
24695 extract_cond_operands.
24697 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24699 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
24700 (extract_cond_operands): Detect condition comparing against non-
24701 invariant bound and return appropriate enum value.
24702 (find_interesting_uses_cond): Update use of extract_cond_operands.
24703 Handle its return value accordingly.
24704 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
24706 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24708 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
24709 nonlinear iv_use computation in loop invariant sensitive way.
24711 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24713 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
24714 (find_iv_candidates): Call relate_compare_use_with_all_cands.
24716 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24718 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
24719 (dump_cand): Support iv_cand.inv_exprs.
24720 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
24722 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
24725 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24727 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
24729 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
24730 as local function. Include necessary header files.
24731 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
24733 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24735 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
24737 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24739 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24740 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24741 RSHIFT_EXPR and BIT_NOT_EXPR.
24743 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24745 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24746 (adjust_setup_cost): New parameter supporting round up adjustment.
24747 (struct address_cost_data): Delete.
24748 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24749 (split_address_cost, ptr_difference_cost): Delete.
24750 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24751 (struct ainc_cost_data): New struct.
24752 (get_address_cost_ainc): New function.
24753 (get_address_cost, get_computation_cost): Reimplement.
24754 (determine_group_iv_cost_address): Record inv_expr for all uses of
24756 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24757 (iv_ca_has_deps): Reimplemented to ...
24758 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24760 (iv_ca_extend): Call iv_ca_more_deps.
24762 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24764 * tree-ssa-address.c (struct mem_address): Move to header file.
24765 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24766 * tree-ssa-address.h (struct mem_address): Move from C file.
24767 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
24769 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24771 * tree-affine.h (aff_combination_type): New interface.
24772 (aff_combination_zero_p): Remove static.
24773 (aff_combination_const_p): New interface.
24774 (aff_combination_singleton_var_p): New interfaces.
24776 2017-05-11 Richard Biener <rguenther@suse.de>
24778 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24779 Skip unreachable blocks and destinations.
24780 (eliminate): Move stmt removal and fixup ...
24781 (fini_eliminate): ... here. Skip inserted exprs.
24782 (pass_pre::execute): Move fini_pre after fini_eliminate.
24783 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
24784 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
24785 PRE to get rid of dead code that has invalid SSA form and
24786 split critical edges again.
24788 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24790 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24792 2017-05-11 Richard Biener <rguenther@suse.de>
24794 * passes.c (execute_function_todo): Verify loops if they are
24795 said to be up-to-date.
24796 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
24797 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
24799 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
24802 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
24803 handle calling assemble_external ourself.
24806 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
24807 modes with zero size. Enhance comment.
24809 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24811 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
24812 built-ins for vec_xl and vec_xst with short and char pointer
24815 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24817 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
24818 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24819 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24820 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24821 (_mm_maskz_min_round_ss): New intrinsics.
24822 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24823 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24824 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24825 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24826 (__builtin_ia32_minss_mask_round): New builtins.
24827 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24828 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24829 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24831 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24832 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24834 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24837 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24839 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24840 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24841 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24842 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24843 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24844 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24845 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24846 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24847 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24848 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24849 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24850 (__builtin_ia32_mulss_mask_round): New builtins.
24851 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24852 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24853 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24855 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24856 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24858 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24861 2017-05-10 Julia Koval <julia.koval@intel.com>
24863 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24864 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24865 (_mm256_setr_m128i): New intrinsics.
24867 2017-05-10 Julia Koval <julia.koval@intel.com>
24869 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24870 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24871 (_mm_maskz_rcp14_ss): New intrinsics.
24872 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24873 (__builtin_ia32_rcp14ss_mask): New builtins.
24874 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24876 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24878 PR tree-optimization/51513
24879 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24880 (assert_unreachable_fallthru_edge_p): Use it.
24881 (group_case_labels_stmt): Likewise.
24882 * tree-cfg.h: Prototype it.
24883 * stmt.c: Include cfghooks.h and tree-cfg.h.
24884 (emit_case_dispatch_table) <gap_label>: New local variable.
24885 Use it to fill dispatch table gaps.
24886 Test for default_label before updating probabilities.
24887 (expand_case) <default_label>: Remove unneeded initialization.
24888 Test for unreachable default case statement and remove its edge.
24889 Set default_label accordingly.
24890 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24892 2017-05-10 Carl Love <cel@us.ibm.com>
24894 * config/rs6000/rs6000-c: Add support for built-in functions
24895 vector signed char vec_neg (vector signed char)
24896 vector signed short int vec_neg (vector short int)
24897 vector signed int vec_neg (vector signed int)
24898 vector signed long long vec_neg (vector signed long long)
24899 vector float vec_neg (vector float)
24900 vector double vec_neg (vector double)
24901 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24903 * config/rs6000/altivec.h: Add define for vec_neg
24904 * doc/extend.texi: Update the built-in documentation for the
24905 new built-in functions.
24907 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24909 PR tree-optimization/77644
24910 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24912 2017-05-10 Nathan Sidwell <nathan@acm.org>
24914 * dumpfile.h (TDI_lang_all): New.
24915 (TDF_KIND): New. Renumber others
24916 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24918 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24920 (get_dump_file_name): Adjust suffix generation.
24921 (dump_enable_all): Use TDF_KIND.
24922 * doc/invoke.texi (-fdump-lang-all): Document.
24924 * dumpfile.h: Tabify.
24926 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24929 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24930 Move member access before delete.
24932 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24934 * tree-inline.c (expand_call_inline): Split block at stmt
24937 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24940 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24941 are now unused after splitting mov{sf,sd}_hardfloat.
24942 (f32_lr2): Likewise.
24943 (f32_lm): Likewise.
24944 (f32_lm2): Likewise.
24945 (f32_li): Likewise.
24946 (f32_li2): Likewise.
24947 (f32_lv): Likewise.
24948 (f32_sr): Likewise.
24949 (f32_sr2): Likewise.
24950 (f32_sm): Likewise.
24951 (f32_sm2): Likewise.
24952 (f32_si): Likewise.
24953 (f32_si2): Likewise.
24954 (f32_sv): Likewise.
24955 (f32_dm): Likewise.
24956 (f32_vsx): Likewise.
24957 (f32_av): Likewise.
24958 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24959 For movsf, order stores so the VSX stores occur before the GPR
24960 store which encourages the register allocator to use a traditional
24961 FPR instead of a GPR. For movsd, order the stores so that the GPR
24962 store comes before the VSX stores to allow the power6 to work.
24963 This is due to the power6 not having a 32-bit integer store
24964 instruction from a FPR.
24965 (movsf_hardfloat): Likewise.
24966 (movsd_hardfloat): Likewise.
24968 2017-05-09 Martin Sebor <msebor@redhat.com>
24970 PR translation/80280
24971 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24974 PR translation/80280
24975 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24976 data member added in r247778.
24977 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24979 2017-05-09 Nathan Sidwell <nathan@acm.org>
24981 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24983 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24986 2017-05-09 Marek Polacek <polacek@redhat.com>
24988 * doc/invoke.texi: Fix typo.
24990 2017-05-09 Richard Biener <rguenther@suse.de>
24992 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24993 (vrp_val_is_min): Likewise.
24994 (set_value_range_to_value): Likewise.
24995 (set_value_range_to_nonnegative): Likewise.
24996 (gimple_assign_nonzero_p): Likewise.
24997 (gimple_stmt_nonzero_p): Likewise.
24998 (vrp_int_const_binop): Likewise. Remove unreachable case.
24999 (adjust_range_with_scev): Adjust comments.
25000 (compare_range_with_value): Likewise.
25001 (extract_range_from_phi_node): Likewise.
25002 (test_for_singularity): Likewise.
25004 2017-05-09 Richard Biener <rguenther@suse.de>
25006 * tree-vrp.c (get_single_symbol): Add assert that we don't
25007 get overflowed constants as invariant part.
25008 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25009 checks. Use wi::cmp instead of recursing for integer constants.
25010 (compare_values): Just ignore whether we assumed undefined
25011 overflow instead of failing the compare.
25012 (extract_range_for_var_from_comparison_expr): Add comment before the
25013 TREE_NO_WARNING sets.
25014 (test_for_singularity): Likewise.
25015 (extract_range_from_comparison): Do not disable optimization
25016 when we assumed undefined overflow.
25017 (extract_range_basic): Remove init of unused var.
25019 2017-05-09 Richard Biener <rguenther@suse.de>
25021 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25022 (extract_range_from_multiplicative_op_1): Adjust.
25023 (extract_range_from_binary_expr_1): Use int_const_binop.
25025 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25028 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25029 rs6000_store_data_bypass_p in seven define_bypass directives and
25030 in several comments.
25031 * config/rs6000/rs6000-protos.h: Add prototype for
25032 rs6000_store_data_bypass_p function.
25033 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25034 function implements slightly different (rs6000-specific) semantics
25035 than store_data_bypass_p, returning false rather than aborting
25036 with assertion error when arguments do not satisfy the
25037 requirements of store data bypass.
25038 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25039 rs6000_store_data_bypass_p.
25041 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25043 * config/xtensa/xtensa-protos.h
25044 (xtensa_initial_elimination_offset): New declaration.
25045 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25046 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25047 macro definition, add case for FRAME_POINTER_REGNUM when
25048 FRAME_GROWS_DOWNWARD.
25049 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25050 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25051 xtensa_initial_elimination_offset.
25053 2017-05-08 Nathan Sidwell <nathan@acm.org>
25055 * doc/invoke.texi: Alphabetize -fdump options.
25057 2017-05-08 Martin Sebor <msebor@redhat.com>
25059 PR translation/80280
25060 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25062 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25064 * target.def (compute_frame_layout): New optional target hook.
25065 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25066 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25067 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25069 * reload1.c (verify_initial_elim_offsets): Likewise.
25070 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25071 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25072 (arm_get_frame_offsets): Split up into this ...
25073 (arm_compute_frame_layout): ... and this function.
25075 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25077 * config/aarch64/constraints.md (Usa): New constraint.
25078 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25080 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25082 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25083 with_multilib_list after it has been checked.
25085 2017-05-08 Richard Biener <rguenther@suse.de>
25087 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25088 (bitmap_set_subtract_values): Likewise.
25090 2017-05-08 Richard Biener <rguenther@suse.de>
25092 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25093 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25095 (gimple_stmt_nonzero_warnv_p): Rename to ...
25096 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25098 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25099 (extract_range_basic): Adjust, do not disable propagation on
25100 strict overflow sensitive simplification.
25101 (vrp_visit_cond_stmt): Likewise.
25103 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25105 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25106 body size unconditionally.
25108 2017-05-07 Jeff Law <law@redhat.com>
25111 2017-05-06 Jeff Law <law@redhat.com>
25112 PR tree-optimization/78496
25113 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25116 PR tree-optimization/78496
25117 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25118 (simplify_stmt_using_ranges): Call it.
25119 (vrp_dom_walker::before_dom_children): Extract equivalences
25120 from an ASSERT_EXPR with an equality comparison against a
25123 2017-05-06 Jeff Law <law@redhat.com>
25125 PR tree-optimization/78496
25126 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25129 PR tree-optimization/78496
25130 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25131 (simplify_stmt_using_ranges): Call it.
25132 (vrp_dom_walker::before_dom_children): Extract equivalences
25133 from an ASSERT_EXPR with an equality comparison against a
25136 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25138 * lra-constraints.c (lra_copy_reg_equiv): New function.
25139 (split_reg): Use it to copy equivalence information from the
25140 original register to the spill register.
25142 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25144 PR rtl-optimization/75964
25145 * simplify-rtx.c (simplify_const_relational_operation): Remove
25146 invalid handling of comparisons of integer ABS.
25148 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25150 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25151 initialize to zero.
25152 (init_regs): Remove declaration.
25153 (function_arg_advance_32): Initialize error_p as boolean variable.
25155 2017-05-05 Nathan Sidwell <nathan@acm.org>
25157 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25158 lines. Use for (;;).
25160 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25162 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25163 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25164 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25165 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25166 VF=2 that require versioning.
25168 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25170 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25173 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25175 * diagnostic.h (diagnostic_override_option_index): Convert from
25176 macro to inline function.
25178 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25180 * diagnostic.c (last_module_changed_p): New function.
25181 (set_last_module): New function.
25182 (diagnostic_report_current_module): Convert macro usage to
25183 the above functions.
25184 * diagnostic.h (diagnostic_context::last_module): Strengthen
25185 from const line_map * to const line_map_ordinary *.
25186 (diagnostic_last_module_changed): Delete macro.
25187 (diagnostic_set_last_module): Delete macro.
25189 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25191 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25192 with diagnostic_report_diagnostic.
25193 (diagnostic_n_impl_richloc): Likewise.
25194 * diagnostic.h (report_diagnostic): Delete macro.
25195 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25196 with diagnostic_report_diagnostic.
25197 * substring-locations.c (format_warning_va): Likewise.
25199 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25201 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25202 save/restor of format_spec. Move option-printing code to...
25203 (print_option_information): ...this new function, and
25204 reimplement by simply printing to the pretty_printer,
25205 rather than appending to the format string.
25207 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25209 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25210 handling logic into...
25211 (update_effective_level_from_pragmas): ...this new function.
25213 2017-05-04 Andrew Waterman <andrew@sifive.com>
25215 * config/riscv/riscv.opt (mstrict-align): New option.
25216 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25217 (SLOW_UNALIGNED_ACCESS): Define.
25218 (riscv_slow_unaligned_access): Declare.
25219 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25221 (riscv_slow_unaligned_access): New variable.
25222 (rocket_tune_info): Set slow_unaligned_access to true.
25223 (optimize_size_tune_info): Set slow_unaligned_access to false.
25224 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25225 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25226 (riscv_option_override): Set riscv_slow_unaligned_access.
25227 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25229 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25231 * config/riscv/riscv.md: Unify indentation.
25233 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25238 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25240 (extendsi<mode>2): Add support for doing sign extension via
25241 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25242 don't have ISA 3.0 instructions.
25243 (extendsi<mode>2 splitter): Likewise.
25244 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25245 generate the normal insns since SImode can now go in vector
25246 registers. Disallow the special UNSPECs needed for previous
25247 machines to hide SImode being used. Add new insns
25248 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25249 (fix_trunc<mode>si2_stfiwx): Likewise.
25250 (fix_trunc<mode>si2_internal): Likewise.
25251 (fixuns_trunc<mode>si2): Likewise.
25252 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25253 (fctiw<u>z_<mode>_smallint): Likewise.
25254 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25255 of floating point to 32-bit integer from doing a direct move to
25256 the GPR registers to do a store.
25257 (fctiwz_<mode>): Break long line.
25259 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25261 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25262 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25263 (addr_list, addr_offset_valid_p): New.
25264 (split_address_groups): Check offset validity with above function.
25265 (gt-tree-ssa-loop-ivopts.h): Include header file.
25267 2017-05-05 Nathan Sidwell <nathan@acm.org>
25269 * config.gcc (arm*-*-*): Add missing 'fi'.
25271 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25273 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25274 included in -fopt-info does not matter.
25275 * doc/optinfo.texi (-fopt-info): Fix description of default
25276 behavour. Explicitly say order of options included in -fopt-info
25279 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25281 * config.gcc: Allow combinations of aprofile and rmprofile values for
25282 --with-multilib-list.
25283 * config/arm/t-multilib: New file.
25284 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25285 variables. Remove setting of ISA and floating-point ABI in
25286 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25287 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25288 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25289 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25291 * config/arm/t-rmprofile: Likewise except for the matches changes.
25292 * doc/install.texi (--with-multilib-list): Document the combination of
25293 aprofile and rmprofile values and warn about pitfalls in doing that.
25295 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25297 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25298 (movdi_aarch64): Likewise.
25300 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25302 PR tree-optimization/80632
25303 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25305 (build_arrays): Initialize it for virtual phis.
25306 (fix_phi_nodes): Use it for virtual phis.
25308 PR tree-optimization/80558
25309 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25310 [x, y] op z into [x op, y op z] for op & or | if conditions
25313 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25314 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25317 * config/arm/arm.md (use_literal_pool): Remove.
25318 (64-bit immediate split): No longer takes cost into consideration
25319 if arm_disable_literal_pool is enabled.
25320 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25321 used when arm_disable_literal_pool is enabled.
25322 (arm_max_const_double_inline_cost): Remove use of
25323 arm_disable_literal_pool.
25324 (push_minipool_fix): Add assert.
25325 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25326 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25327 (no_literal_pool_sf_immediate): New.
25329 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25331 PR tree-optimization/80613
25332 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25333 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25335 2017-05-05 Richard Biener <rguenther@suse.de>
25337 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25339 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25341 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25342 of this flag from insn conditions due to removal from r247495.
25344 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25346 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25348 (arm_early_store_addr_dep_ptr): Likewise.
25349 * config/arm/aarch-common-protos.h
25350 (arm_early_load_addr_dep_ptr): Add prototype.
25351 (arm_early_store_addr_dep_ptr): Likewise.
25352 * config/arm/cortex-a53.md: Add new bypasses.
25354 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25356 * tree.c (next_type_uid): Change type to unsigned.
25357 (type_hash_canon): Decrement back next_type_uid if
25358 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25359 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25362 2017-05-04 Martin Sebor <msebor@redhat.com>
25364 * builtins.c: Fix a trivial typo in a comment.
25366 PR middle-end/79234
25367 * builtins.c (check_sizes): Adjust to handle reading past the end.
25368 Avoid printing excessive upper bound of ranges. Use %E to print
25369 tree nodes instead of converting them to %wu.
25370 (expand_builtin_memchr): New function.
25371 (compute_dest_size): Rename...
25372 (compute_objsize): ...to this.
25373 (expand_builtin_memcpy): Adjust.
25374 (expand_builtin_mempcpy): Adjust.
25375 (expand_builtin_strcat): Adjust.
25376 (expand_builtin_strcpy): Adjust.
25377 (check_strncat_sizes): Adjust.
25378 (expand_builtin_strncat): Adjust.
25379 (expand_builtin_strncpy): Adjust and simplify.
25380 (expand_builtin_memset): Adjust.
25381 (expand_builtin_bzero): Adjust.
25382 (expand_builtin_memcmp): Adjust.
25383 (expand_builtin): Handle memcmp.
25384 (maybe_emit_chk_warning): Check strncat just once.
25386 2017-05-04 Martin Sebor <msebor@redhat.com>
25388 PR preprocessor/79214
25389 PR middle-end/79222
25390 PR middle-end/79223
25391 * builtins.c (check_sizes): Add inlining context and issue
25392 warnings even when -Wno-system-headers is set.
25393 (check_strncat_sizes): Same.
25394 (expand_builtin_strncat): Same.
25395 (expand_builtin_memmove): New function.
25396 (expand_builtin_stpncpy): Same.
25397 (expand_builtin): Handle memmove and stpncpy.
25399 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25401 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25402 which is not used any more.
25404 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25406 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25408 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25410 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25411 (cortexa53_tunings): Likewise.
25412 (cortexa57_tunings): Likewise.
25413 (cortexa72_tunings): Likewise.
25414 (cortexa73_tunings): Likewise.
25416 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25418 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25419 Set loop alignment to 8.
25421 2017-05-04 Martin Sebor <msebor@redhat.com>
25423 PR translation/80280
25424 * builtins.c (expand_builtin_object_size): Add missing quoting to
25425 %D and like directives.
25426 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25427 (hsa_type_for_tree_type): Same.
25428 (verify_function_arguments): Same.
25429 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25430 * varasm.c (get_section): Same.
25433 2017-05-04 Martin Sebor <msebor@redhat.com>
25435 PR translation/80280
25436 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25438 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25440 * config/aarch64/aarch64.c (generic_addrcost_table):
25441 Change HI/TI mode setting.
25443 2017-05-04 Martin Jambor <mjambor@suse.cz>
25445 PR tree-optimization/80622
25446 * tree-sra.c (comes_initialized_p): New function.
25447 (build_accesses_from_assign): Only set write lazily when
25448 comes_initialized_p is false.
25449 (analyze_access_subtree): Use comes_initialized_p.
25450 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25451 instead of testing for PARM_DECL.
25453 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25455 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25456 constraint on operand 0 to allow more general addressing modes.
25457 Adjust output template.
25458 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25460 * config/aarch64/aarch64-protos.h
25461 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25462 * config/aarch64/constraints.md (Dp): New address constraint.
25463 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25466 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25468 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25469 update use of estimate_ipcp_clone_size_and_time.
25470 (estimate_local_effects): Update use of
25471 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25472 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25473 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25474 Return nonspecialized time.
25476 2017-05-04 Richard Biener <rguenther@suse.de>
25478 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25479 for the last VUSE which def dominates the PHI. Directly call
25481 (get_continuation_for_phi_1): Remove.
25483 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25485 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25486 to explain the use of truncating division. Cap the number of
25487 iterations to the maximum given by nb_iterations_upper_bound,
25490 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25492 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25493 * configure: Regenerate.
25494 * config.in: Regenerate.
25495 * config/i386/driver-mingw32.c: new file.
25496 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25497 * config.host: Link driver-mingw32.o on MinGW host.
25498 * doc/install.texi: Document new --enable-mingw-wildcard configure
25501 2017-05-04 Marek Polacek <polacek@redhat.com>
25503 PR tree-optimization/80612
25504 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25506 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25507 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25509 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25510 (movt splitter): Likewise.
25511 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25512 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25513 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25514 block for Thumb-1 with MOVT.
25515 (thumb2_legitimate_address_p): Move code block ...
25516 (can_avoid_literal_pool_for_label_p): ... into this new function.
25517 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25519 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25520 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25521 "M-profile targets with the MOVT instruction".
25523 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25525 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25526 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25527 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25529 2017-05-04 Martin Liska <mliska@suse.cz>
25531 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25532 variable cond_code.
25534 2017-05-04 Richard Biener <rguenther@suse.de>
25536 * tree.c (array_at_struct_end_p): Handle arrays at struct
25537 end with flexarrays more conservatively. Refactor and treat
25538 arrays of arrays or aggregates more strict. Fix
25539 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25540 * tree.c (array_at_struct_end_p): Adjust prototype.
25541 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25542 * gimple-fold.c (get_range_strlen): Likewise.
25543 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25545 2017-05-04 Richard Biener <rguenther@suse.de>
25547 PR tree-optimization/31130
25548 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25550 (supports_overflow_infinity): Likewise.
25551 (is_negative_overflow_infinity): Likewise.
25552 (is_positive_overflow_infinity): Likewise.
25553 (is_overflow_infinity): Likewise.
25554 (stmt_overflow_infinity): Likewise.
25555 (overflow_infinity_range_p): Likewise.
25556 (usable_range_p): Remove as always returning true.
25557 (make_overflow_infinity): Remove.
25558 (negative_overflow_infinity): Likewise.
25559 (positive_overflow_infinity): Likewise.
25560 (avoid_overflow_infinity): Likewise.
25561 (set_value_range): Adjust accordingly.
25562 (set_value_range_to_nonnegative): Likewise, remove now unused
25563 overflow_infinity arg.
25564 (vrp_operand_equal_p): Adjust.
25565 (update_value_range): Likewise.
25566 (range_int_cst_singleton_p): Likewise.
25567 (operand_less_p): Likewise.
25568 (compare_values_warnv): Likewise.
25569 (extract_range_for_var_from_comparison_expr): Likewise.
25570 (vrp_int_const_binop): Likewise.
25571 (zero_nonzero_bits_from_vr): Likewise.
25572 (extract_range_from_multiplicative_op_1): Likewise.
25573 (extract_range_from_binary_expr_1): Likewise.
25574 (extract_range_from_unary_expr): Likewise.
25575 (extract_range_from_comparison): Likewise.
25576 (extract_range_basic): Likewise.
25577 (adjust_range_with_scev): Likewise.
25578 (compare_ranges): Likewise.
25579 (compare_range_with_value): Likewise.
25580 (dump_value_range): Likewise.
25581 (test_for_singularity): Likewise, remove strict_overflow_p parameter
25583 (simplify_cond_using_ranges): Adjust.
25585 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25587 * brig-builtins.def: Added a builtin for class_f64.
25588 * builtin-types.def: Added a builtin type needed by class_f64.
25590 2017-05-03 Jason Merrill <jason@redhat.com>
25592 * timevar.def: Add TV_CONSTEXPR.
25594 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25596 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
25598 2017-05-03 Martin Jambor <mjambor@suse.cz>
25600 * ipa-prop.c (ipa_update_after_lto_read): Removed.
25601 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
25602 * ipa-cp.c (ipcp_propagate_stage): Do not call
25603 ipa_update_after_lto_read.
25604 * ipa-inline.c (ipa_inline): Likewise.
25606 2017-05-03 Martin Jambor <mjambor@suse.cz>
25608 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
25609 tag. Added a default constructor and a destructor.
25610 (ipa_edge_args_sum_t): New class;
25611 (ipa_edge_args_sum): Declare.
25612 (ipa_edge_args_vector): Remove declaration.
25613 (IPA_EDGE_REF): Use ipa_edge_args_sum.
25614 (ipa_free_edge_args_substructures): Remove declaration.
25615 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
25616 (ipa_edge_args_info_available_for_edge_p): Likewise.
25617 * ipa-prop.c (ipa_edge_args_vector): Removed.
25618 (edge_removal_hook_holder): Likewise.
25619 (edge_duplication_hook_holder): Likewise.
25620 (ipa_edge_args_sum): New variable.
25621 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
25622 ipa_edge_args_vector.
25623 (ipa_free_edge_args_substructures): Likewise.
25624 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
25625 ipa_edge_args_vector.
25626 (ipa_edge_removal_hook): Turned into method
25627 ipa_edge_args_sum_t::remove.
25628 (ipa_edge_duplication_hook): Turned into method
25629 ipa_edge_args_sum_t::duplicate.
25630 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
25631 registering edge hooks.
25632 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
25633 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
25634 ipa_edge_args_sum instead of ipa_edge_args_vector.
25635 * ipa-profile.c (ipa_profile): Likewise.
25637 2017-05-03 Martin Jambor <mjambor@suse.cz>
25639 * symbol-summary.h (function_summary): New method exists.
25640 (function_summary::symtab_removal): Deallocate through release.
25641 (call_summary): New class.
25642 (gt_ggc_mx): New overload.
25643 (gt_pch_nx): Likewise.
25644 (gt_pch_nx): Likewise.
25646 2017-05-03 Jeff Law <law@redhat.com>
25648 PR tree-optimization/78496
25649 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
25650 from simplify_cond_using_ranges. Split off code to walk
25651 backwards through casts into ...
25652 (simplify_cond_using_ranges_2): New function.
25653 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
25654 (execute_vrp): After identifying jump threads, call
25655 simplify_cond_using_ranges_2.
25657 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
25660 * ipa-inline.h (inline_summary): Add ctor.
25661 (create_ggc): Do not use ggc_cleared_alloc.
25663 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
25664 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25666 * gcc.c (handle_braces): Support escaping in switch matching
25668 * doc/invoke.texi (Spec Files): Document it.
25669 Remove superfluous @code markup in items.
25671 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25673 * diagnostic-show-locus.c (struct column_range): New struct.
25674 (get_affected_columns): New function.
25675 (get_printed_columns): New function.
25676 (struct correction): New struct.
25677 (correction::ensure_capacity): New function.
25678 (correction::ensure_terminated): New function.
25679 (struct line_corrections): New struct.
25680 (line_corrections::~line_corrections): New dtor.
25681 (line_corrections::add_hint): New function.
25682 (layout::print_trailing_fixits): Reimplement in terms of the new
25684 (selftest::test_overlapped_fixit_printing): New function.
25685 (selftest::diagnostic_show_locus_c_tests): Call it.
25687 2017-05-03 Nathan Sidwell <nathan@acm.org>
25689 Canonicalize canonical type hashing
25690 * tree.h (type_hash_canon_hash): Declare.
25691 * tree.c (type_hash_list, attribute_hash_list): Move into
25692 type_hash_canon_hash.
25693 (build_type_attribute_qual_variant): Break out hash code calc into
25694 type_hash_canon_hash.
25695 (type_hash_canon_hash): New. Generic type hash computation.
25696 (build_range_type_1, build_array_type_1, build_function_type,
25697 build_method_type_directly, build_offset_type, build_complex_type,
25698 make_vector_type): Call it.
25700 2017-05-03 Richard Biener <rguenther@suse.de>
25702 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
25703 When all DRs have unknown misaligned do not always peel
25704 when there is a store but apply the same costing model as if
25705 there were only loads.
25707 2017-05-03 Richard Biener <rguenther@suse.de>
25710 PR tree-optimization/80492
25711 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25712 compare_base_decls returning dont-know properly.
25714 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
25716 * config/arm/iterators.md (CCSI): New mode iterator.
25717 (arch): New mode attribute.
25718 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
25719 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
25720 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
25721 code iterator for success result mode.
25722 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
25723 the corresponding new insn generators.
25725 2017-05-03 Bin Cheng <bin.cheng@arm.com>
25728 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25729 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25731 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
25733 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
25734 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
25735 (DDR_A): Wrap DDR argument in brackets.
25736 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
25737 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
25738 (DDR_REVERSED_P): Likewise.
25740 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25742 PR tree-optimization/79472
25743 * tree-switch-conversion.c (struct switch_conv_info): Add
25744 contiguous_range and default_case_nonstandard fields.
25745 (collect_switch_conv_info): Compute contiguous_range and
25746 default_case_nonstandard fields, don't clear final_bb if
25747 contiguous_range and only the default case doesn't have the required
25749 (check_all_empty_except_final): Set default_case_nonstandard instead
25750 of failing if contiguous_range and the default case doesn't have empty
25752 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25753 and only the default case doesn't have the required constants. Skip
25755 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25756 if default_case_nonstandard.
25757 (build_constructors): Build constant 1 just once. Assert that default
25758 values aren't inserted in between cases if contiguous_range. Skip
25760 (build_arrays): Skip virtual phis.
25761 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25762 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25763 Handle virtual phis.
25764 (gen_inbound_check): Handle default_case_nonstandard case.
25765 (process_switch): Adjust check_final_bb caller. Call
25766 gather_default_values with the first non-default case instead of
25767 default case if default_case_nonstandard.
25769 2017-05-02 Nathan Sidwell <nathan@acm.org>
25771 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
25772 check. Fix formatting.
25774 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
25776 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
25777 errors when comparing specialized and unspecialized times.
25779 2017-05-02 David Malcolm <dmalcolm@redhat.com>
25781 * diagnostic-show-locus.c
25782 (layout::should_print_annotation_line_p): Make private.
25783 (layout::print_annotation_line): Make private.
25784 (layout::annotation_line_showed_range_p): Make private.
25785 (layout::show_ruler): Make private.
25786 (layout::print_source_line): Make private. Pass in line and
25787 line_width, rather than calling location_get_source_line. Drop
25789 (layout::print_leading_fixits): New method.
25790 (layout::print_any_fixits): Rename to...
25791 (layout::print_trailing_fixits): ...this, and make private.
25792 Don't print newline fixits.
25793 (diagnostic_show_locus): Move logic for printing one row into...
25794 (layout::print_line): ...this new function. Move the
25795 location_get_source_line call and error-handling from
25796 print_source_line to here. Call print_leading_fixits, and rename
25797 print_any_fixits to print_trailing_fixits.
25798 (selftest::test_fixit_insert_containing_newline): Update now that
25799 newlines are partially supported.
25800 (selftest::test_fixit_insert_containing_newline_2): New test.
25801 (selftest::test_fixit_replace_containing_newline): Update comments.
25802 (selftest::diagnostic_show_locus_c_tests): Call the new test.
25803 * edit-context.c (class added_line): New class.
25804 (class edited_line): Describe newline handling in comment.
25805 (edited_line::actually_edited_p): New method.
25806 (edited_line::print_content): Delete redundant decl.
25807 (edited_line::m_predecessors): New field.
25808 (edited_file::print_content): Call edited_line::print_content.
25809 (edited_file::print_diff): Update to support newlines.
25810 (edited_file::print_diff_hunk): Likewise.
25811 (edited_file::print_run_of_changed_lines): New function.
25812 (edited_file::print_diff_line): Convert to...
25813 (print_diff_line): ...this.
25814 (edited_file::get_effective_line_count): New function.
25815 (edited_line::edited_line): Initialize new field m_predecessors.
25816 (edited_line::~edited_line): Clean up m_predecessors.
25817 (edited_line::apply_fixit): Handle newlines.
25818 (edited_line::get_effective_line_count): New function.
25819 (edited_line::print_content): New function.
25820 (edited_line::print_diff_lines): New function.
25821 (selftest::test_applying_fixits_insert_containing_newline): New
25823 (selftest::test_applying_fixits_replace_containing_newline): New
25825 (selftest::insert_line): New function.
25826 (selftest::test_applying_fixits_multiple_lines): Add example of
25828 (selftest::edit_context_c_tests): Call the new tests.
25830 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25832 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25833 parameter cand. Update dump information.
25834 (get_computation_cost): Update uses.
25836 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25838 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25839 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25840 (get_computation_at, rewrite_use_address): Update use of
25841 get_computation_aff.
25843 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25845 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25846 (get_computation): Delete.
25847 (get_computation_cost): Implement like get_computation_cost_at.
25848 Use get_computation_at.
25849 (get_computation_cost_at): Delete.
25850 (rewrite_use_nonlinear_expr): Use get_computation_at.
25851 (rewrite_use_compare, remove_unused_ivs): Ditto.
25853 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25855 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25857 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25859 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25860 (ivopts_global_cost_for_size): Rename parameter and update uses.
25861 (iv_ca_recount_cost): Update uses.
25862 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25863 candidates seperately in n_invs and n_cands.
25864 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25866 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25868 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25869 (find_inv_vars_cb): New.
25870 (find_depends): Renamed to ...
25871 (find_inv_vars): ... this.
25872 (add_candidate_1, force_var_cost): Call find_inv_vars.
25873 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25875 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25877 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25878 inv_vars. Add inv_exprs.
25879 (struct iv_cand): Rename depends_on to inv_vars.
25880 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25881 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25882 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25883 (dump_cand): Dump inv_vars.
25884 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25885 (record_invariant, find_depends, add_candidate_1): Ditto.
25886 (set_group_iv_cost, force_var_cost): Ditto.
25887 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25888 (get_computation_cost_at, get_computation_cost): Ditto.
25889 (determine_group_iv_cost_generic): Ditto.
25890 (determine_group_iv_cost_address): Ditto.
25891 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25892 (determine_group_iv_costs): Ditto.
25893 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25894 (iv_ca_set_remove_invariants): Renamed to ...
25895 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25896 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25897 (iv_ca_set_add_invariants): Renamed to ...
25898 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25899 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25900 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25901 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25902 (create_new_ivs): Remove useless dump.
25904 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25906 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25908 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25909 (iv_ca_set_no_cp, create_new_iv): Ditto.
25911 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25913 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25915 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25917 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25918 function tree_check2.
25920 2017-05-02 Martin Liska <mliska@suse.cz>
25922 * doc/gcov.texi: Add missing preposition.
25923 * gcov.c (function_info::function_info): Properly fill up
25924 all member variables.
25926 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25928 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25930 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25932 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25934 2017-05-02 Martin Liska <mliska@suse.cz>
25937 * lto-streamer-in.c (lto_read_tree_1): Remove
25938 LTO_STREAMER_DEBUG.
25939 * lto-streamer.c (struct tree_hash_entry): Likewise.
25940 (struct tree_entry_hasher): Likewise.
25941 (tree_entry_hasher::hash): Likewise.
25942 (tree_entry_hasher::equal): Likewise.
25943 (lto_streamer_init): Likewise.
25944 (lto_orig_address_map): Likewise.
25945 (lto_orig_address_get): Likewise.
25946 (lto_orig_address_remove): Likewise.
25947 * lto-streamer.h: Likewise.
25948 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25949 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25951 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25953 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25954 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25955 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25956 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25957 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25958 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25959 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25960 (mm_maskz_sub_ss): New intrinsics.
25961 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25962 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25963 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25964 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25965 (__builtin_ia32_subss_mask_round): New builtins.
25966 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25967 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25968 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25970 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25971 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25973 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25976 2017-05-02 Martin Jambor <mjambor@suse.cz>
25978 PR tree-optimization/78687
25979 * tree-sra.c (access): New field parent.
25980 (process_subtree_disqualification): New function.
25981 (disqualify_candidate): Call it.
25982 (build_accesses_from_assign): Reset write flag if creating an
25984 (build_access_subtree): Fill in parent field and also prpagate
25985 down grp_write flag.
25986 (create_artificial_child_access): New parameter set_grp_write, set
25987 grp_write to its value.
25988 (propagate_subaccesses_across_link): Also propagate grp_write flag
25990 (propagate_all_subaccesses): Push the closest parent back to work
25991 queue if add_access_to_work_queue returned true.
25993 2017-05-02 Richard Biener <rguenther@suse.de>
25995 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25996 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25997 -fstrict-overflow documentation.
25998 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25999 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26000 flag_strict_overflow.
26001 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26002 * lto-opts.c (lto_write_options): Do not stream it.
26003 * lto-wrapper.c (merge_and_complain): Do not handle it.
26004 * opts.c (default_options_table): Do not set -fstrict-overflow.
26005 (finish_options): Likewise do not clear it when sanitizing.
26006 * simplify-rtx.c (simplify_const_relational_operation): Do not
26007 test flag_strict_overflow.
26009 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26011 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26012 using enabled attribute.
26013 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26014 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26015 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26016 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26017 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26018 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26019 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26020 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26021 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26022 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26024 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26026 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26028 2017-05-02 Richard Biener <rguenther@suse.de>
26030 PR tree-optimization/80591
26032 2017-04-10 Richard Biener <rguenther@suse.de>
26034 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26037 2017-05-02 Richard Biener <rguenther@suse.de>
26039 PR tree-optimization/80549
26040 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26041 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26042 headers if they do not have a preheader.
26044 2017-05-02 Martin Liska <mliska@suse.cz>
26047 * common.opt: Fix typo.
26048 * doc/invoke.texi: Likewise.
26050 2017-05-01 Jan Beulich <jbeulich@suse.com>
26052 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26053 swapping, add (x,x,m,x,n) alternative.
26055 2017-05-01 Nathan Sidwell <nathan@acm.org>
26057 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26058 unnecessary unadjusted_alignment check.
26060 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26063 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26065 * gimplify.c (gimplify_cilk_detach): New function.
26066 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26067 * tree-core.h: Document EXPR_CILK_SPAWN.
26068 * tree.h (EXPR_CILK_SPAWN): Define.
26070 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26072 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26073 to use new fixit_hint representation, using the "replace" logic.
26074 (get_line_span_for_fixit_hint): Likewise.
26075 (layout::print_any_fixits): Likewise.
26076 (selftest::test_one_liner_many_fixits): Rename to...
26077 (selftest::test_one_liner_many_fixits_1): ...this, and update
26078 comment and expected output to reflect that the multiple fix-it
26079 hints are now consolidated into one insertion.
26080 (selftest::test_one_liner_many_fixits_2): New test.
26081 (selftest::test_diagnostic_show_locus_one_liner): Update for
26083 (selftest::test_fixit_consolidation): Update for fix-it API
26085 * diagnostic.c (print_parseable_fixits): Likewise.
26086 * edit-context.c (edited_line::m_line_events): Convert from
26087 auto_vec <line_event *> to auto_vec <line_event>.
26088 (class line_event): Convert from abstract base class to a concrete
26089 class, taking over the role of replace_event.
26090 (class insert_event): Delete.
26091 (class replace_event): Rename to class line_event. Convert to
26093 (edit_context::add_fixits): Reimplement.
26094 (edit_context::apply_insert): Delete.
26095 (edit_context::apply_replace): Rename to...
26096 (edit_context::apply_fixit): ...this. Convert to half-open range.
26097 (edited_file::apply_insert): Delete.
26098 (edited_file::apply_replace): Rename to...
26099 (edited_file::apply_fixit): ...this.
26100 (edited_line::~edited_line): Drop deletion of events.
26101 (edited_line::apply_insert): Delete.
26102 (edited_line::apply_replace): Rename to...
26103 (edited_line::apply_fixit): ...this. Convert to half-open range.
26104 Update for change to type of m_line_events.
26105 * edit-context.h (edit_context::apply_insert): Delete.
26106 (edit_context::apply_replace): Rename to...
26107 (edit_context::apply_fixit): ...this.
26109 2017-05-01 Martin Sebor <msebor@redhat.com>
26111 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26114 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26117 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26118 __get_cpuid_max returns 0.
26119 (__get_cpuid_count): Ditto.
26121 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26123 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26124 replacement expression is another instance of one of its arguments.
26126 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26129 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26130 check for stack push/pop autoinc.
26131 * config/i386/i386.c (ix86_agi_dependent): Return false
26132 if the only reason why modified_in_p returned true is that
26133 addr is SP based and set_insn is a push or pop.
26135 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26137 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26140 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26143 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26144 (account_size_time): Use two predicates - exec_pred and
26146 (evaluate_conditions_for_known_args): Compute both clause and
26148 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26149 (inline_summary_t::duplicate): Update.
26150 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26152 (compute_inline_parameters): Likewise.
26153 (estimate_edge_size_and_time): Update caluclation of time.
26154 (estimate_node_size_and_time): Compute both time and nonspecialized
26156 (estimate_ipcp_clone_size_and_time): Update.
26157 (inline_merge_summary): Update.
26158 (do_estimate_edge_time): Update.
26159 (do_estimate_edge_size): Update.
26160 (do_estimate_edge_hints): Update.
26161 (inline_read_section, inline_write_summary): Stream both new predicates.
26162 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26164 (compute_inlined_call_time): Cleanup.
26165 (big_speedup_p): Update.
26166 (edge_badness): Update.
26167 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26168 (size_time_entry): Replace predicate by exec_predicate and
26169 nonconst_predicate.
26170 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26171 (estimate_edge_time): Return also nonspec_time.
26172 (reset_edge_growth_cache): Update.
26174 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26176 PR rtl-optimization/80491
26177 * ifcvt.c (noce_process_if_block): When looking for x setter
26178 with missing else_bb, don't check only the insn right before
26179 cond_earliest, but look for the last insn that x is modified in
26180 within the same bb.
26182 PR rtl-optimization/80491
26183 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26185 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26187 PR tree-optimization/80487
26188 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26190 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26192 PR tree-optimization/79697
26193 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26194 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26195 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26197 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26198 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26200 2017-04-28 Martin Sebor <msebor@redhat.com>
26202 PR tree-optimization/80523
26203 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26204 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26206 (maybe_warn, format_directive, parse_directive): Use new functions.
26207 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26209 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26211 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26213 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26215 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26216 target_header_dir): Set correctly.
26217 * configure: Regenerated.
26218 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26219 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26220 instead of SYSTEM_HEADER_DIR.
26222 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26224 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26225 (estimate_local_effects): Likewise.
26226 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26227 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26228 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26229 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26230 do_estimate_edge_time, estimate_edge_time): Likewise.
26231 * ipa-inline-analysis.c (estimate_node_size_and_time,
26232 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26233 (estimate_time_after_inlining): Remove.
26235 2017-04-28 Martin Liska <mliska@suse.cz>
26237 * doc/gcov.texi: Enhance documentation of gcov.
26239 2017-04-28 Martin Liska <mliska@suse.cz>
26241 * doc/gcov.texi: Sort options in alphabetic order.
26242 * doc/gcov-dump.texi: Likewise.
26243 * doc/gcov-tool.texi: Likewise.
26244 * gcov.c (print_usage): Likewise.
26245 * gcov-dump.c (print_usage): Likewise.
26246 * gcov-tool.c (print_merge_usage_message): Likewise.
26247 (print_rewrite_usage_message): Likewise.
26248 (print_overlap_usage_message): Likewise.
26250 2017-04-28 Martin Liska <mliska@suse.cz>
26252 PR gcov-profile/53915
26253 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26255 2017-04-28 Martin Liska <mliska@suse.cz>
26257 PR gcov-profile/79891
26258 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26259 is marked by compiler as living on a line.
26260 (get_cycles_count): Remove usage of the union.
26261 (output_intermediate_file): Likewise.
26262 (find_source): Fix GNU coding style.
26263 (accumulate_line_counts): Remove old non-all block mode.
26264 (output_lines): Remove usage of the union.
26265 * profile.c (output_location): Include all BBs, even if
26266 belonging to a same line (and file) as a previous BB.
26268 2017-04-28 Martin Liska <mliska@suse.cz>
26270 * gcov.c (process_args): Handle new argument 'w'.
26271 (read_graph_file): Assign ID to BBs.
26272 (output_branch_count): Display BB # if verbose flag is set.
26273 (output_lines): Likewise for arcs.
26274 (print_usage): Add '--verbose' option help.
26275 * doc/gcov.texi: Document --verbose (-w) option.
26277 2017-04-28 Martin Liska <mliska@suse.cz>
26279 * gcov.c (struct block_location_info): New struct.
26280 (process_file): Fill up the new structure.
26281 (read_graph_file): Replace usage of encoding by the newly added
26283 (add_line_counts): Likewise.
26284 (accumulate_line_counts): Remove usage of the union.
26285 (function_info::function_info): New function.
26286 (function_info::~function_info): Likewise.
26287 (process_file): Call delete instead of release_function.
26288 (release_function): Release the function.
26289 (release_structures): Call delete instead of release_function.
26290 (solve_flow_graph): Replace usage of num_blocks.
26291 (find_exception_blocks): Likewise.
26292 (output_lines): Fix GNU coding style.
26294 2017-04-28 Martin Liska <mliska@suse.cz>
26297 * coverage.c (coverage_remove_note_file): New function.
26298 * coverage.h: Declare the function.
26299 * toplev.c (finalize): Clean if an error has been seen.
26301 2017-04-28 Martin Liska <mliska@suse.cz>
26303 PR gcov-profile/80031
26304 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26305 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26306 * gcov.c (read_graph_file): Read just number of blocks.
26307 * profile.c (branch_prob): Do not stream 0 flags per a basic
26310 2017-04-28 Martin Liska <mliska@suse.cz>
26312 * gcov-dump.c (tag_*): Add new argument to declarations.
26313 (dump_gcov_file): Likewise.
26314 (tag_blocks): Add and use new argument depth.
26315 (tag_arcs): Likewise.
26316 (tag_lines): Likewise.
26317 (tag_counters): Likewise.
26318 (tag_summary): Likewise.
26319 (dump_working_sets): Use depth to do a proper indentation.
26321 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26324 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26325 * symtab.c (symtab_node::debug_symtab): Move definition here.
26327 2017-04-28 Richard Biener <rguenther@suse.de>
26329 * lto-streamer.h (LTO_major_version): Bump to 7.
26331 2017-04-28 Richard Biener <rguenther@suse.de>
26333 * tree-vrp.c (assert_info): New struct.
26334 (add_assert_info): New helper.
26335 (register_edge_assert_for_2): Refactor to add asserts to a vector
26337 (register_edge_assert_for_1): Likewise.
26338 (register_edge_assert_for): Likewise.
26339 (finish_register_edge_assert_for): New helper actually registering
26340 asserts where live on edge.
26341 (find_conditional_asserts): Adjust.
26342 (find_switch_asserts): Likewise.
26343 (evrp_dom_walker::try_find_new_range): Generalize.
26344 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26346 2017-04-27 Marek Polacek <polacek@redhat.com>
26349 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26350 arg10 and arg11 to itype.
26352 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26354 * doc/extend.texi (Object Size Checking): Improve grammar.
26356 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26359 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26360 that the logic for permitting reciprocal estimates matches that
26363 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26366 * tree.c (type_cache_hasher::equal): Only compare
26367 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26368 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26369 non-aggregate element types.
26370 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26371 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26373 2017-04-27 Richard Biener <rguenther@suse.de>
26375 PR middle-end/80533
26376 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26377 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26378 keeping a reference to a trailing array.
26380 2017-04-27 Richard Biener <rguenther@suse.de>
26382 PR middle-end/80539
26383 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26384 being in loop-closed SSA form conservatively.
26385 (chrec_fold_multiply_poly_poly): Likewise.
26387 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26389 PR middle-end/79665
26390 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26391 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26393 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26396 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26397 (aarch64_function_arg_alignment): Return unsigned int again, but still
26398 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26399 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26400 Don't emit -Wpsabi note.
26401 (aarch64_function_arg_boundary): Likewise.
26402 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26405 2017-04-26 Nathan Sidwell <nathan@acm.org>
26407 * tree.h (crc32_unsigned_n): Declare.
26408 (crc32_unsigned, crc32_unsigned): Make inline.
26409 * tree.c (crc32_unsigned_bits): Replace with ...
26410 (crc32_unsigned_n): ... this.
26411 (crc32_unsigned, crc32_byte): Remove.
26412 (crc32_string): Remove unnecessary braces.
26414 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26416 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26417 * ipa-inline-analysis.c (MAX_TIME): Remove.
26418 (account_size_time): Use sreal for time.
26419 (dump_inline_summary): Update.
26420 (estimate_function_body_sizes): Update.
26421 (estimate_edge_size_and_time): Update.
26422 (estimate_calls_size_and_time): Update.
26423 (estimate_node_size_and_time): Update.
26424 (inline_merge_summary): Update.
26425 (inline_update_overall_summary): Update.
26426 (estimate_time_after_inlining): Update.
26427 (inline_read_section): Update.
26428 (inline_write_summary): Update.
26429 * ipa-inline.c (compute_uninlined_call_time): Update.
26430 (compute_inlined_call_time): Update.
26431 (recursive_inlining): Update.
26432 (inline_small_functions): Update.
26433 (dump_overall_stats): Update.
26434 * ipa-inline.h: Include sreal.h.
26435 (size_time_entry): Turn time to sreal.
26436 (inline_summary): Turn self_time nad time to sreal.
26438 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26440 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26442 (sreal::stream_out, sreal::stream_in): New.
26443 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26445 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26447 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26450 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26453 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26455 (dimode_scalar_chain::compute_convert_gain): Ditto.
26456 (dimode_scalar_chain::make_vector_copies): Ditto.
26457 (dimode_scalar_chain::convert_reg): Ditto.
26458 (dimode_scalar_chain::convert_insn): Ditto.
26459 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26460 (VI248_AVX512BW_1): New mode iterator.
26461 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26462 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26465 2017-04-25 Martin Sebor <msebor@redhat.com>
26467 PR tree-optimization/80497
26468 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26469 constants are representable in HOST_WIDE_INT.
26470 (parse_directive): Ditto.
26472 2017-04-25 Martin Sebor <msebor@redhat.com>
26475 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26476 (new_zero_array): Adjust signature.
26477 (dom_info::dom_init): Used unsigned rather that size_t.
26478 (dom_info::dom_info): Same.
26480 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26481 Jakub Jelinek <jakub@redhat.com>
26484 * config/arm/arm.c: Include gimple.h.
26485 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26486 returns negative, increment ncrn only if it returned positive.
26487 (arm_needs_doubleword_align): Return int instead of bool,
26488 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26489 members, but if there is any such non-FIELD_DECL
26490 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26491 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26492 returns negative, increment nregs only if it returned positive.
26493 (arm_setup_incoming_varargs): Likewise.
26494 (arm_function_arg_boundary): Emit -Wpsabi note if
26495 arm_needs_doubleword_align returns negative, return
26496 DOUBLEWORD_ALIGNMENT only if it returned positive.
26498 2017-04-25 Marek Polacek <polacek@redhat.com>
26501 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26502 first argument to type.
26504 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26507 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26508 type checks to test for compatibility instead of equality.
26510 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26511 Jakub Jelinek <jakub@redhat.com>
26514 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26516 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26517 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26518 the alignment computation, but return their maximum in warn_alignment.
26519 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26520 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26522 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26523 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26526 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26528 * config/arc/simdext.md (dmpyh): Fix typo.
26530 2017-04-25 Richard Biener <rguenther@suse.de>
26532 PR tree-optimization/80492
26533 * alias.c (compare_base_decls): Handle registers with asm
26534 specification conservatively.
26535 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26536 compare_base_decls returning dont-know properly.
26538 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26540 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26541 (legitimate_offset_address_p): New function.
26542 (arc_legitimate_address_p): Use above function.
26544 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26546 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26548 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26550 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26551 ACCH registers whenever they are available.
26553 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26555 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
26556 double regs fix when not used.
26558 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26560 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
26562 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
26563 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
26565 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26567 * config/arc/arc.c (arc_output_addsi): Check for h-register class
26568 when emitting short ADD instructions.
26570 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26572 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
26574 (cmpsi_cc_c_insn): Likewise.
26575 (cbranchsi4_scratch): Compute proper instruction length using
26576 compact_hreg_operand.
26577 * config/arc/predicates.md (compact_hreg_operand): New predicate.
26579 2017-04-25 Richard Biener <rguenther@suse.de>
26581 PR middle-end/80509
26582 * passes.c (pass_manager::pass_manager): Initialize
26583 m_name_to_pass_map.
26585 2017-04-25 Richard Biener <rguenther@suse.de>
26587 PR tree-optimization/79201
26588 * tree-ssa-sink.c (statement_sink_location): Handle calls.
26590 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26593 * config/s390/vector.md: Split MEM->GPR vector moves for
26594 non-s_operand addresses.
26596 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26599 * config/s390/predicates.md (reload_const_wide_int_operand): New
26601 * config/s390/s390.md ("movti"): Remove d/P alternative.
26602 ("movti_bigconst"): New pattern definition.
26604 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26607 * s390-protos.h (s390_expand_cs_hqi): Removed.
26608 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
26609 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
26610 modes as well as CCZ1mode and CCZmode.
26611 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
26612 signature of s390_emit_compare_and_swap.
26613 (s390_expand_cs_hqi): Likewise, make static.
26614 (s390_expand_cs_tdsi): Generate an explicit compare before trying
26615 compare-and-swap, in some cases.
26616 (s390_expand_cs): Wrapper function.
26617 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
26619 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
26620 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
26621 patterns for small and large integers. Forbid symref memory operands.
26622 Move expander to s390.c. Require cc register.
26623 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
26624 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
26625 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
26626 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
26627 symref memory operands. Remove CC mode and call s390_match_ccmode
26629 ("atomic_exchange<mode>"): Allow and implement all integer modes.
26631 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26633 * config/s390/s390.md (define_peephole2): New peephole to help
26634 combining the load-and-test pattern with volatile memory.
26636 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26638 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
26639 with CCZmode for TARGET_Z196.
26641 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26643 PR rtl-optimization/80501
26644 * combine.c (make_compound_operation_int): Set subreg_code to SET
26645 even for AND with mask of the sign bit of mode.
26647 PR rtl-optimization/80500
26648 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
26649 sum's initial value.
26651 2017-04-25 Julian Brown <julian@codesourcery.com>
26652 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26654 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
26656 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
26658 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
26660 2017-04-25 Julian Brown <julian@codesourcery.com>
26661 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26663 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
26664 (thunderx2t99_sha): New Reservation.
26666 2017-04-25 Julian Brown <julian@codesourcery.com>
26667 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26669 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
26670 type for 1-element load.
26672 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
26674 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
26676 2017-04-24 Martin Jambor <mjambor@suse.cz>
26678 PR tree-optimization/80293
26679 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
26680 char arrays not totally scalarizable if it is false.
26681 (analyze_all_variable_accesses): Pass correct value in the new
26682 parameter. Add a statistics counter.
26684 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
26686 PR middle-end/79931
26687 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
26689 2017-04-24 Richard Biener <rguenther@suse.de>
26691 PR tree-optimization/80494
26692 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
26693 out for complex types.
26695 2017-04-24 Richard Biener <rguenther@suse.de>
26697 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
26698 * tree-ssa-sccvn.c (print_scc): Print SCC size.
26699 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
26700 (DFS): Adjust and never fail.
26701 (sccvn_dom_walker::fail): Remove.
26702 (sccvn_dom_walker::before_dom_children): Adjust.
26703 (run_scc_vn): Likewise and never fail.
26704 * tree-ssa-pre.c (pass_pre::execute): Adjust.
26705 (pass_fre::execute): Likewise.
26707 2017-04-24 Richard Biener <rguenther@suse.de>
26709 PR tree-optimization/79725
26710 * tree-ssa-sink.c (statement_sink_location): Return whether
26711 failure reason was zero uses. Move that check later.
26712 (sink_code_in_bb): Deal with zero uses by removing the stmt
26715 2017-04-24 Richard Biener <rguenther@suse.de>
26718 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
26719 pointer-based references.
26721 2017-04-24 Richard Biener <rguenther@suse.de>
26724 * pass_manager.h (pass_manager::operator new): Remove.
26725 (pass_manager::operator delete): Likewise.
26726 * passes.c (pass_manager::operator new): Remove.
26727 (pass_manager::operator delete): Likewise.
26728 (pass_manager::pass_manager): Zero individual pass members.
26730 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
26733 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
26734 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
26735 Check "XEXP (src, 1)" operand here.
26736 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
26737 Check "XEXP (src, 1)" operand here.
26738 (dimode_scalar_chain::make_vector_copies): Detect count register
26739 of a shift instruction. Zero extend count register from QImode
26740 to DImode to satisfy vector shift pattern count operand predicate.
26741 Substitute vector shift count operand with a DImode copy.
26742 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26745 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26747 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26748 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26749 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26750 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26751 (UNSPEC_NOREX_MEM): Remove definition.
26753 2017-04-21 Richard Biener <rguenther@suse.de>
26755 PR tree-optimization/79547
26756 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26757 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26758 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26759 without any constraints.
26761 2017-04-21 Richard Biener <rguenther@suse.de>
26763 PR tree-optimization/78847
26764 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26766 2017-04-21 Richard Biener <rguenther@suse.de>
26768 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
26769 (build_distinct_type_copy): Likewise.
26770 (build_variant_type_copy): Likewise.
26771 * tree.c (build_qualified_type): Pass down mem-stat info.
26772 (build_distinct_type_copy): Likewise.
26773 (build_variant_type_copy): Likewise.
26775 2017-04-21 Richard Biener <rguenther@suse.de>
26777 PR tree-optimization/80237
26778 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
26780 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
26781 for a simplified result.
26783 2016-04-21 Richard Biener <rguenther@suse.de>
26785 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
26786 sth as strict as a simple_iv but a chrec without symbols and an
26787 operand defined in the loop we are peeling (and not some subloop).
26788 (propagate_constants_for_unrolling): Propagate all constants.
26790 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26793 * config/i386/i386.c (print_reg): Remove assert for disalowed
26794 regno values, call output_operand_lossage instead.
26796 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26799 * config/i386/constraints.md (Yc): New register constraint.
26800 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26801 Use Yc constraint for alternative 2 of operand 0. Remove
26802 preferred_for_speed attribute.
26804 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26806 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
26807 lastprivate clauses in SIMT case.
26809 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
26811 * doc/invoke.texi (-Wextra-semi): Document new warning option.
26813 2017-04-20 Richard Biener <rguenther@suse.de>
26815 PR tree-optimization/57796
26816 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
26817 as N scalar stores.
26818 (vect_model_load_cost): Cost gathers as N scalar loads.
26820 2017-04-20 Richard Biener <rguenther@suse.de>
26822 * ggc-page.c (ggc_allocated_p): Rename to ...
26823 (safe_lookup_page_table_entry): ... this and return the lookup
26825 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26827 2017-04-20 Richard Biener <rguenther@suse.de>
26829 PR tree-optimization/80453
26830 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26831 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26832 from the conditions.
26833 (vn_phi_eq): Pass them down.
26834 (vn_phi_lookup): Record them.
26835 (vn_phi_insert): Likewise.
26837 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26839 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26840 uninitialized variable warning to avoid buffer overrun.
26842 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26845 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26846 is suppressed for '{ 0 }' in C.
26848 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26850 * BASE-VER: Set to 8.0.0.
26852 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26854 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26855 priority .init_array and .fini_array section with SECTION_NOTYPE
26858 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26860 PR middle-end/80423
26861 * tree.h (build_array_type): Add typeless_storage default argument.
26862 * tree.c (type_cache_hasher::equal): Also compare
26863 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26864 (build_array_type): Add typeless_storage argument, set
26865 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26867 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26868 (build_array_type): Likewise. Add typeless_storage argument.
26870 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26871 Jakub Jelinek <jakub@redhat.com>
26873 PR tree-optimization/80426
26874 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26875 operation on symbolic operands, also compute the overflow for the
26876 invariant part when the operation degenerates into a negation.
26878 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26881 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26882 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26885 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26887 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26890 * config/avr/avr.c (tree.h): Include it.
26891 (cgraph.h): Include it.
26892 (avr_encode_section_info): Don't warn for uninitialized progmem
26893 variable if it's just an alias.
26895 2017-04-19 Richard Biener <rguenther@suse.de>
26898 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26899 when needed by AutoPGO.
26901 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26904 * doc/lto.texi: Remove an extra 'that'.
26906 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26908 PR rtl-optimization/80429
26909 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26910 are only used in debug insns.
26912 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26913 Vladimir Makarov <vmakarov@redhat.com>
26915 * config/sparc/predicates.md (input_operand): Add comment. Return
26916 true for any memory operand when LRA is in progress.
26917 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26919 2017-04-18 Jeff Law <law@redhat.com>
26922 * mips.md ({return,simple_return}_internal): Do not overwrite
26925 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26927 PR tree-optimization/80443
26928 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26929 instead of adding 1, subtract -1 and similarly instead of subtracting
26932 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26934 PR rtl-optimization/80357
26935 * haifa-sched.c (tmp_bitmap): New variable.
26936 (model_recompute): Handle duplicate use records.
26937 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26938 (free_global_sched_pressure_data): Free it.
26940 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26943 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26944 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26945 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26946 instead of SYSTEM_HEADER_DIR.
26948 2017-04-18 Jeff Law <law@redhat.com>
26950 PR middle-end/80422
26951 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26952 predecessors after walking up the insn chain.
26954 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26957 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26958 sizetype type into debug info.
26960 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26963 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26964 unneeded test for TARGET_UPPER_REGS_SF.
26965 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26967 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26970 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26971 instead of gsi_after_labels.
26973 2017-04-18 Jeff Law <law@redhat.com>
26975 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26979 2017-04-13 Jeff Law <law@redhat.com>
26980 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26981 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26983 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26986 * config/avr/avr.c (intl.h): Include it.
26987 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26989 2017-04-18 Martin Liska <mliska@suse.cz>
26991 PR gcov-profile/78783
26992 * gcov-tool.c (gcov_output_files): Validate that destination
26993 file is either removed by the tool or by a user.
26995 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26996 Guy Benyei <guybe@mellanox.com>
26998 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26999 block, and do not negate it, the stored id is already negative.
27001 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27003 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27005 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27008 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27009 masks of options that should be turned off if the VSX vector
27010 options are turned off.
27011 (OTHER_P8_VECTOR_MASKS): Likewise.
27012 (OTHER_VSX_VECTOR_MASKS): Likewise.
27013 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27014 rs6000_disable_incompatible_switches to validate no type switches
27016 (rs6000_incompatible_switch): New function to disallow turning on
27017 other vector options if -mno-vsx, -mno-power8-vector, or
27018 -mno-power9-vector are specified.
27020 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27022 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27024 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27026 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27027 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27028 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27029 (ARG_POINTER_CFA_OFFSET): Likewise.
27031 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27033 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27034 conditions to take advantage of various optimizations.
27036 2017-04-13 Jeff Law <law@redhat.com>
27038 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27039 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27040 (zero_extendsidi2_dext): Likewise.
27042 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27045 * fold-const.c (fold_ternary_loc): Revert
27046 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27049 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27051 PR rtl-optimization/80343
27052 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27053 new scratch pseudo.
27055 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27058 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27059 to ubsan_encode_value.
27061 2017-04-13 Jeff Law <law@redhat.com>
27063 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27064 appearing in DEBUG_INSNs.
27066 2017-04-13 Martin Liska <mliska@suse.cz>
27068 PR gcov-profile/80413
27069 * gcov-io.c (gcov_write_string): Copy to buffer just when
27070 allocated size is greater than zero.
27072 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27075 * dwarf2out.c (decls_for_scope): Ignore declarations of
27076 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27078 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27081 * ipa-visibility.c (non_local_p): Fix typos.
27082 (localize_node): When localizing symbol in same comdat group,
27083 dissolve the group only when we know external symbols are going
27085 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27087 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27089 PR tree-optimization/79390
27090 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27091 order does not result in usable sequence, retry with reversed operand
27097 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27098 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27099 op0 instead of fold_convert_loc (loc, type, arg0).
27101 2017-04-12 Jeff Law <law@redhat.com>
27103 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27104 has a delay slot in the generated code.
27106 * config/cris/cris.md (cris_preferred_reload_class): Return
27107 GENNONACR_REGS rather than GENERAL_REGS.
27109 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27112 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27113 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27114 signedness of the result type.
27116 2017-04-12 Richard Biener <rguenther@suse.de>
27117 Jeff Law <law@redhat.com>
27119 PR tree-optimization/80359
27120 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27121 trim stores to TARGET_MEM_REFs.
27123 2017-04-12 Richard Biener <rguenther@suse.de>
27125 PR tree-optimization/79390
27126 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27127 threading case even more.
27129 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27132 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27133 for quad_address_p for TImode, instead of just not indexed_address.
27135 2017-04-12 Richard Biener <rguenther@suse.de>
27136 Bernd Edlinger <bernd.edlinger@hotmail.de>
27138 PR middle-end/79671
27139 * alias.c (component_uses_parent_alias_set_from): Handle
27140 TYPE_TYPELESS_STORAGE.
27141 (get_alias_set): Likewise.
27142 * tree-core.h (tree_type_common): Add typeless_storage flag.
27143 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27144 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27145 for types containing members with TYPE_TYPELESS_STORAGE.
27146 (place_field): Likewise.
27147 (layout_type): Likewise for ARRAY_TYPE.
27148 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27149 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27150 TYPE_TYPELESS_STORAGE.
27151 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27153 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27156 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27157 first argument to type.
27159 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27163 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27164 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27165 (rs6000_expand_binop_builtin): Likewise.
27166 (rs6000_expand_ternop_builtin): Likewise; also add missing
27167 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27168 vshasigma built-ins.
27169 * doc/extend.texi: Document that vec_xxpermdi's third argument
27170 must be a constant.
27172 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27174 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27175 Use shift_const cost parameter when calculating gain of STV shifts.
27177 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27179 PR rtl-optimization/70478
27180 * lra-constraints.c (process_alt_operands): Check memory for
27181 disfavoring memory insn operand.
27183 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27185 PR middle-end/80100
27186 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27187 left shift in unsigned HOST_WIDE_INT type.
27189 PR rtl-optimization/80385
27190 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27191 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27194 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27195 if they have any depend clauses.
27197 2017-04-11 Martin Liska <mliska@suse.cz>
27200 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27201 * ipa-split.c (split_function): Create a local comdat symbol
27202 if caller is in a comdat group.
27204 2017-04-11 Martin Liska <mliska@suse.cz>
27207 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27210 2017-04-11 Martin Sebor <msebor@redhat.com>
27212 PR middle-end/80364
27213 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27214 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27216 (directive::set_width, directive::set_precision, format_character):
27218 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27221 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27224 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27225 conflict, set target->arch_name instead of target->cpu_name.
27227 2017-04-11 Richard Biener <rguenther@suse.de>
27229 PR tree-optimization/80374
27230 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27231 build_zero_cst, remove fold_convertible_p check again.
27233 2017-04-11 Martin Liska <mliska@suse.cz>
27236 * ubsan.c (instrument_object_size): Do not instrument register
27239 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27242 * config/i386/i386-builtin-types.def
27243 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27244 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27245 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27246 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27247 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27248 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27249 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27250 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27251 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27252 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27253 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27254 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27255 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27256 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27257 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27258 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27259 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27260 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27261 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27262 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27263 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27264 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27265 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27266 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27267 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27268 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27269 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27270 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27271 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27272 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27273 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27274 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27275 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27276 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27277 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27278 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27279 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27280 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27281 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27282 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27283 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27284 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27285 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27286 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27287 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27288 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27290 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27291 flag to second_arg_count, handle 4 argument function type _COUNT
27292 aliases, handle second_arg_count on second argument rather than last.
27294 2017-04-10 Jeff Law <law@redhat.com>
27296 PR tree-optimization/80374
27297 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27298 record anything if we can not convert integer_zero_node to the
27301 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27304 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27305 Enhance special handling given to the TARGET_P9_MINMAX option in
27306 relation to certain other options.
27308 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27310 PR tree-optimization/80153
27311 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27312 remove POINTER_PLUS_EXPR's base part directly, rather than through
27315 2017-04-10 Richard Biener <rguenther@suse.de>
27316 Bin Cheng <bin.cheng@arm.com>
27318 PR tree-optimization/80153
27319 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27320 the first element of pointer type aff_tree. Build result expr in
27322 (add_elt_to_tree): Convert to type unconditionally. Remove other
27323 fold_convert calls.
27324 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27325 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27327 2017-04-10 Richard Biener <rguenther@suse.de>
27329 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27332 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27334 PR rtl-optimization/70478
27335 * lra-constraints.c (curr_small_class_check): New.
27336 (update_and_check_small_class_inputs): New.
27337 (process_alt_operands): Update curr_small_class_check. Disfavor
27338 alternative insn memory operands. Check available regs for small
27341 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27344 * config/mips/mips.opt (-mvirt): Update description.
27345 * doc/invoke.texi (-mvirt): Likewise.
27347 2017-04-10 Richard Biener <rguenther@suse.de>
27349 PR middle-end/80362
27350 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27351 looking for NEGATE_EXPR in -A / -B to A / B folding.
27353 2017-04-10 Martin Liska <mliska@suse.cz>
27355 PR gcov-profile/80224
27356 * gcov.c (print_usage): Fix usage string.
27357 (get_gcov_intermediate_filename): Remove.
27358 (output_gcov_file): Use both for normal and intermediate format.
27359 (generate_results): Do not initialize special file for
27360 intermediate format.
27362 2017-04-10 Richard Biener <rguenther@suse.de>
27364 PR tree-optimization/80304
27365 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27368 2017-04-10 Nathan Sidwell <nathan@acm.org>
27371 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27372 (rs6000_init_builtins): Use it.
27374 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27376 * config/arm/arm.md (<mrc>): Add mode to SET source.
27377 (<mrrc>): Likewise.
27379 2017-04-10 Richard Biener <rguenther@suse.de>
27381 PR middle-end/80344
27382 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27384 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27387 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27388 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27389 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27390 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27391 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27392 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27393 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27394 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27395 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27396 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27397 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27398 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27399 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27400 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27401 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27402 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27403 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27404 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27405 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27406 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27407 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27408 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27409 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27411 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27413 PR rtl-optimization/70478
27414 * lra-constraints.c: Reverse the last patch.
27416 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27418 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27419 Add comment for WCHAR_T.
27421 2017-04-08 Martin Liska <mliska@suse.cz>
27424 2017-04-07 Martin Liska <mliska@suse.cz>
27427 * ipa-split.c (split_function): Add function part to a same comdat
27430 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27433 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27435 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27437 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27438 (vec_pairing): ...to this.
27439 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27440 (rs6000_sched_init): Adjust for name change.
27441 (struct rs6000_sched_context): Likewise.
27442 (rs6000_init_sched_context): Likewise.
27443 (rs6000_set_sched_context): Likewise.
27445 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27451 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27453 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27454 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27455 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27457 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27459 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27461 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27463 PR rtl-optimization/70703
27464 * ira-color.c (update_conflict_hard_regno_costs): Use
27465 int64_t instead of HOST_WIDE_INT.
27467 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27469 PR rtl-optimization/70478
27470 * lra-constraints.c (process_alt_operands): Disfavor alternative
27471 insn memory operands.
27473 2017-04-07 Jeff Law <law@redhat.com>
27475 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27476 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27478 2017-04-07 Martin Liska <mliska@suse.cz>
27481 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27482 Show error message instead of an ICE.
27484 2017-04-07 Martin Liska <mliska@suse.cz>
27487 * ipa-split.c (split_function): Add function part to a same comdat
27490 2017-04-07 Richard Biener <rguenther@suse.de>
27492 PR middle-end/80341
27493 * tree.c (get_unwidened): Also handle ! for_type case for
27495 * convert.c (do_narrow): Split out from ...
27496 (convert_to_integer_1): ... here. Do not pass final truncation
27497 type to get_unwidened for TRUNC_DIV_EXPR.
27499 2017-04-07 Richard Biener <rguenther@suse.de>
27501 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27503 (aff_combination_const): Adjust.
27504 (aff_combination_scale): Likewise.
27505 (aff_combination_add_elt): Likewise.
27506 (aff_combination_add_cst): Likewise.
27507 (aff_combination_convert): Likewise.
27508 (add_elt_to_tree): Likewise. Remove unused argument.
27509 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27511 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27513 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27515 * config/arm/arm.c (arm_default_short_enums): Use
27516 ARM_DEFAULT_SHORT_ENUMS.
27517 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27519 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27522 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27523 members with redundant out-of-class redeclaration.
27525 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27528 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27529 * config/i386/i386.md (*zero_extendsidi2):
27530 Add (?*x,*x) and (?*v,*v) alternatives.
27532 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27535 * config/i386/i386.c (ix86_expand_builtin)
27536 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27537 mode from insn data. Convert operands to insn operand mode.
27538 Copy operands that don't satisfy insn predicate to a register.
27540 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27542 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27545 2017-04-06 Richard Biener <rguenther@suse.de>
27547 PR tree-optimization/80334
27548 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27549 preserve alignment of accesses.
27551 2017-04-06 Richard Biener <rguenther@suse.de>
27553 PR tree-optimization/80262
27554 * tree-sra.c (build_ref_for_offset): Preserve address-space
27556 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
27557 Drop useless address-space information on MEM_REF offsets.
27559 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
27561 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
27563 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27565 PR rtl-optimization/70703
27566 * ira-color.c (update_conflict_hard_regno_costs): Use
27567 HOST_WIDE_INT instead of long.
27569 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
27572 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
27573 not defined for x86_64 target. Add -mmmx target option when __SSE2__
27575 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
27576 for x86_64 target. Handle -m3dnowa option.
27578 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27580 PR rtl-optimization/70703
27581 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
27582 (update_conflict_hard_regno_costs): Use long instead of unsigned
27583 arithmetic for cost calculation.
27585 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27586 Bernd Edlinger <bernd.edlinger@hotmail.de>
27589 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
27592 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
27595 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
27596 ptr_mode with Pmode throughout.
27597 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
27598 into probe_stack_range and use DImode.
27600 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
27603 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
27604 call_eh_return is true.
27606 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27608 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
27609 Initialize last_match_fntype_index.
27611 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27614 * tree-nvr.c: Include internal-fn.h.
27615 (pass_return_slot::execute): Ignore internal calls without
27618 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27619 Richard Biener <rguenther@suse.de>
27622 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
27623 captures used multiple times, except for the last use.
27624 * generic-match-head.c: Include gimplify.h.
27626 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27628 PR tree-optimization/79390
27629 * target.h (struct noce_if_info): Declare.
27630 * targhooks.h (default_noce_conversion_profitable_p): Declare.
27631 * target.def (noce_conversion_profitable_p): New target hook.
27632 * ifcvt.h (struct noce_if_info): New type, moved from ...
27633 * ifcvt.c (struct noce_if_info): ... here.
27634 (noce_conversion_profitable_p): Renamed to ...
27635 (default_noce_conversion_profitable_p): ... this. No longer
27637 (noce_try_store_flag_constants, noce_try_addcc,
27638 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
27639 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
27640 instead of noce_conversion_profitable_p.
27641 * config/i386/i386.c: Include ifcvt.h.
27642 (ix86_option_override_internal): Don't override
27643 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
27644 (ix86_noce_conversion_profitable_p): New function.
27645 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
27646 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
27647 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
27648 * doc/tm.texi: Regenerated.
27650 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27652 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
27655 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27658 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
27659 instructions for small multiply cores.
27661 2017-04-04 Jeff Law <law@redhat.com>
27663 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
27665 (mips_expand_vec_perm_const): Initialize elements in orig_perm
27666 that are not set by the loop over the elements.
27668 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27671 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
27672 int mode, convert_modes it to mode as unsigned, otherwise use
27673 lowpart_subreg to mode rather than SImode.
27674 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
27675 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
27676 Use DImode instead of SImode for the shift count operand.
27677 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
27680 2017-04-04 Richard Biener <rguenther@suse.de>
27682 PR middle-end/80281
27683 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
27684 arithmetic done for the negate or the plus. Simplify.
27685 (A - (-B) -> A + B): Likewise.
27686 * fold-const.c (split_tree): Make sure to not negate pointers.
27688 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
27690 PR rtl-optimization/60818
27691 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
27692 a compare of comparisons with the thing compared if this results
27693 in a different machine mode.
27695 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
27697 * alias.c (base_alias_check): Fix typo in comment.
27698 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
27699 * cgraphunit.c (symbol_table::compile): Likewise.
27700 * collect2.c (maybe_run_lto_and_relink): Likewise.
27701 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
27702 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
27703 * config/avr/avr.c (avr_map_op_t): Likewise.
27704 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
27705 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
27706 * config/epiphany/epiphany.md (movcc): Likewise.
27707 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
27708 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
27710 * config/mips/mips.c (mips_save_restore_reg): Likewise.
27711 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
27712 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
27713 * config/sh/sh.c (sh_rtx_costs): Likewise.
27714 * fold-const.c (fold_truth_andor): Likewise.
27715 * genautomata.c (collapse_flag): Likewise.
27716 * gengtype.h (struct type::u::s): Likewise.
27717 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
27718 * input.c (FORMAT_AMOUNT): Likewise.
27719 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
27720 (known_aggs_to_agg_replacement_list): Likewise.
27721 * ipa-inline-analysis.c: Likewise.
27722 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
27723 * ipa-polymorphic-call.c
27724 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
27725 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
27726 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
27728 * modulo-sched.c (apply_reg_moves): Likewise.
27729 * omp-expand.c (build_omp_regions_1): Likewise.
27730 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
27731 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
27732 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
27733 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27734 * value-prof.c: Likewise.
27735 * var-tracking.c (val_reset): Likewise.
27737 2017-04-03 Richard Biener <rguenther@suse.de>
27739 PR tree-optimization/80275
27740 * fold-const.c (split_address_to_core_and_offset): Handle
27743 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27745 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27746 descriptors is at least equal to that of functions.
27748 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27750 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27752 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27755 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27756 (mov<IMOD4:mode>): New expander.
27757 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27759 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27761 PR rtl-optimization/79405
27762 * fwprop.c (propagations_left): New variable.
27763 (forward_propagate_into): Decrement it.
27764 (fwprop_init): Initialize it.
27765 (fw_prop): If the variable has reached zero, stop propagating.
27766 (fwprop_addr): Ditto.
27768 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27771 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
27772 a FUNCTION_DECL, pass it as decl instead of origin to
27775 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
27777 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
27780 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
27783 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
27784 TARGET_VSX_SMALL_INTEGER.
27786 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27788 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27789 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
27791 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27793 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
27794 extraction from odd-numbered MSA register.
27796 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27798 PR middle-end/80173
27799 * expmed.c (store_bit_field_1): Don't attempt to create
27800 a word subreg out of hard registers wider than word if they
27801 have HARD_REGNO_NREGS of 1 for their mode.
27803 PR middle-end/80163
27804 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
27805 conversions to integer types wider than word and pointer.
27808 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
27809 (rtx_equal_for_cselib_p): Pass 0 to it.
27810 * cselib.c (cselib_hasher::equal): Likewise.
27811 (rtx_equal_for_cselib_1): Add depth argument. If depth
27812 is 128, don't look up VALUE locs and punt. Increment
27813 depth in recursive calls when walking VALUE locs.
27815 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
27817 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
27818 (make_gcov_file_name): Use the canonical path name for generating
27820 (read_line): Fix handling of files with ascii null bytes.
27822 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27824 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27825 to initialise a vector register instead
27826 of using a const_int.
27828 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27830 PR translation/80189
27831 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27832 diagnostic messages.
27834 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27837 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27838 (dfp_diex_<mode>): Update mode of operand 1.
27839 * doc/extend.texi (dxex, dxexq): Document change to return type.
27840 (diex, diexq): Document change to argument type.
27842 2017-03-30 Martin Jambor <mjambor@suse.cz>
27845 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27846 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27847 it reflects the signature changes performed at the callee side.
27848 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27849 to cgraph_build_function_type_skip_args.
27850 (build_function_decl_skip_args): Adjust call to the above function.
27852 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27855 * config/i386/sse.md
27856 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27857 register as dest whenever it is a MEM not rtx_equal_p to the
27858 corresponding dup operand, and when forcing into reg move the
27859 reg into the memory afterwards.
27860 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27861 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27862 for the force_reg mode.
27863 (avx512vl_vextractf128<mode>): Use register as dest either
27864 always when a MEM, or when it is a MEM not rtx_equal_p to the
27865 corresponding dup operand, or even not when it is a CONST_VECTOR
27866 depending on the mode and lo vs. hi.
27867 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27869 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27870 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27871 Likewise. Require that operands[2] is even.
27872 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27873 Remove extraneous parens. Require that operands[2] is a multiple
27875 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27876 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27877 disallow memory then.
27879 2017-03-30 Richard Biener <rguenther@suse.de>
27881 PR tree-optimization/77498
27882 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27883 to non-constants over backedges.
27885 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27887 PR rtl-optimization/80233
27888 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27889 as last_combined_insn. Do not test for BARRIER_P separately.
27891 2017-03-29 Andreas Schwab <schwab@suse.de>
27894 * calls.c (prepare_call_address): Convert funexp to Pmode before
27895 copying to temp reg.
27897 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27899 PR tree-optimization/80158
27900 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27901 Handle possible future case of more than one alternate
27903 (replace_rhs_if_not_dup): Likewise.
27904 (replace_one_candidate): Likewise.
27906 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27908 PR rtl-optimization/80193
27909 * ira.c (ira): Do not check allocation for LRA.
27911 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27913 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27914 (nvptx_output_simt_exit): Declare.
27915 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27916 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27917 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27918 (nvptx_declare_function_name): ...here. Emit declaration of local
27919 memory space buffer for omp_simt_enter insn.
27920 (nvptx_output_unisimt_switch): New.
27921 (nvptx_output_softstack_switch): New.
27922 (nvptx_output_simt_enter): New.
27923 (nvptx_output_simt_exit): New.
27924 * config/nvptx/nvptx.h (struct machine_function): New fields
27925 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27926 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27927 (UNSPECV_SIMT_EXIT): Ditto.
27928 (omp_simt_enter_insn): New insn.
27929 (omp_simt_enter): New expansion.
27930 (omp_simt_exit): New insn.
27931 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27933 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27934 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27935 (expand_GOMP_SIMT_EXIT): New.
27936 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27937 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27938 (GOMP_SIMT_EXIT): Ditto.
27939 * target-insns.def (omp_simt_enter): New insn.
27940 (omp_simt_exit): Ditto.
27941 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27943 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27944 (lower_rec_input_clauses): Likewise.
27945 (lower_lastprivate_clauses): Handle SIMT privatization.
27947 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27948 (ompdevlow_adjust_simt_enter): New.
27949 (find_simtpriv_var_op): New.
27950 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27951 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27953 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27954 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27955 (copy_decl_for_dup_finish): Ditto.
27957 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27959 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27962 * config/i386/i386.c (ix86_option_override_internal): Always
27963 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27965 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27967 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27969 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27971 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27972 mark new edge's irreducible flag accordign to it.
27973 (vect_do_peeling): Check loop preheader edge's irreducible flag
27974 and pass it to function slpeel_add_loop_guard.
27976 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27978 PR tree-optimization/80218
27979 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27980 Update block frequencies and counts.
27982 2017-03-28 Richard Biener <rguenther@suse.de>
27984 PR tree-optimization/78644
27985 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27986 of a simplification result we may not use it at all.
27988 2017-03-28 Richard Biener <rguenther@suse.de>
27991 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27992 without arguments, generate default definition of a SSA name.
27994 2017-03-28 Richard Biener <rguenther@suse.de>
27996 PR middle-end/80222
27997 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27998 TYPE_REF_CAN_ALIAS_ALL references.
27999 * fold-const.c (fold_indirect_ref_1): Likewise.
28001 2017-03-28 Martin Liska <mliska@suse.cz>
28004 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28005 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28007 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28008 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28010 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28011 (EXTRA_SPECS): Define.
28012 (SUBTARGET_EXTRA_SPECS): Likewise.
28013 (SUBTARGET_CPP_SPEC): Likewise.
28014 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28015 SUBTARGET_EXTRA_SPECS.
28016 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28018 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28020 * config/arc/simdext.md (vst64_insn): Update pattern.
28021 (vld32wh_insn): Likewise.
28022 (vld32wl_insn): Likewise.
28023 (vld64_insn): Likewise.
28024 (vld32_insn): Likewise.
28026 2017-03-28 Marek Polacek <polacek@redhat.com>
28029 * fold-const.c (fold_comparison): Use protected_set_expr_location
28030 instead of SET_EXPR_LOCATION.
28032 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28034 * tree.c (add_expr): Avoid name lookup warning.
28036 2017-03-27 Jeff Law <law@redhat.com>
28038 PR tree-optimization/80216
28039 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28040 function name. Limit recursion depth.
28041 (record_temporary_equivalences): Corresponding changes.
28043 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28045 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28048 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28051 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28053 * cfgcleanup.c (reg_note_cfa_p): New array.
28054 (insns_have_identical_cfa_notes): New function.
28055 (old_insns_match_p): Don't cross-jump in between /f
28056 and non-/f instructions. If both i1 and i2 are frame related,
28057 verify all CFA notes, their order and content.
28059 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28062 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28063 HImode and SImode with zero extend to DImode to one insn.
28064 (bswap<mode>2_extenddi): Likewise.
28065 (bswapsi2_extenddi): Likewise.
28066 (bswaphi2_extendsi): Likewise.
28067 (bswaphi2): Combine bswap HImode and SImode into one insn.
28068 Separate memory insns from swapping register.
28069 (bswapsi2): Likewise.
28070 (bswap<mode>2): Likewise.
28071 (bswaphi2_internal): Delete, no longer used.
28072 (bswapsi2_internal): Likewise.
28073 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28074 store, and gpr<-gpr swap insns.
28075 (bswap<mode>2_store): Likewise.
28076 (bswaphi2_reg): Register only splitter, combine with the splitter.
28077 (bswaphi2 splitter): Likewise.
28078 (bswapsi2_reg): Likewise.
28079 (bswapsi2 splitter): Likewise.
28080 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28081 the insns into load, store, and register/register insns.
28082 (bswapdi2_ldbrx): Likewise.
28083 (bswapdi2_load): Likewise.
28084 (bswapdi2_store): Likewise.
28085 (bswapdi2_reg): Likewise.
28087 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28089 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28090 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28092 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28095 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28097 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28098 special handling for target option conflicts between dform
28099 options (-mpower9-dform, -mpower9-dform-vector,
28100 -mpower9-dform-scalar) and -mno-direct-move.
28102 2017-03-27 Richard Biener <rguenther@suse.de>
28104 PR tree-optimization/80181
28105 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28107 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28109 * config/arc/predicates.md (move_double_src_operand): Replace the
28110 call to move_double_src_operand with a call to address_operand.
28112 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28114 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28115 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28116 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28118 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28120 * config/arc/predicates.md (long_immediate_loadstore_operand):
28121 Consider scaled addresses cases.
28123 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28125 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28126 restored when in interrupt.
28127 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28128 doesn't have delay slot.
28130 2017-03-27 Richard Biener <rguenther@suse.de>
28133 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28134 inlined thunk clones.
28136 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28139 * asan.c (instrument_derefs): Copy over last operand from
28140 original COMPONENT_REF to the new COMPONENT_REF with
28141 DECL_BIT_FIELD_REPRESENTATIVE.
28142 * ubsan.c (instrument_object_size): Likewise.
28144 2017-03-27 Richard Biener <rguenther@suse.de>
28146 PR tree-optimization/80170
28147 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28148 sure DR/SCEV didnt fold in constants we do not see when looking
28149 at the reference base alignment.
28151 2017-03-27 Richard Biener <rguenther@suse.de>
28153 PR middle-end/80171
28154 * gimple-fold.c (fold_ctor_reference): Properly guard against
28155 NULL return value from canonicalize_constructor_val.
28157 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28160 * config/i386/i386.c (ix86_expand_builtin)
28161 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28162 flags reg setting and flags reg using instructions.
28163 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28164 clobbering instructions to zero extend op2.
28166 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28168 * doc/install.texi (Configuration) <--with-aix-soname>:
28169 Update link to AIX ld.
28171 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28173 PR rtl-optimization/80160
28174 PR rtl-optimization/80159
28175 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28176 reg_alternate_class into account.
28178 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28181 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28182 to consider in curr_insn_transform.
28184 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28186 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28187 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28188 and emit_mode_inner.
28190 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28192 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28193 argument to the overloaded builtin variants. Use the new flag to
28194 deprecate certain builtin variants.
28195 * config/s390/s390-builtin-types.def: Add new builtin types.
28196 * config/s390/s390-builtins.h: Support new flags field for
28197 overloaded builtins.
28198 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28199 (s390_macro_to_expand): Enable vector float data type.
28200 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28201 builtins by incrementing the __VEC__ version number.
28202 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28204 (s390_resolve_overloaded_builtin): Emit error messages depending
28205 on the builtin flags.
28206 * config/s390/s390.c (s390_expand_builtin): Support additional
28207 flags argument. Change error message to match the messages
28208 emitted in s390-c.c.
28209 * config/s390/s390.md: New UNSPEC_* constants.
28210 (op_type): Add new instruction types.
28211 * config/s390/vecintrin.h: Add new builtins and test data class
28213 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28214 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28215 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28216 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28217 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28218 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28220 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28221 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28222 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28223 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28225 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28226 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28227 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28228 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28230 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28231 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28232 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28233 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28234 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28235 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28236 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28238 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28239 ("vec_scatter_element<V_HW_4:mode>_DI")
28240 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28241 ("vec_fpint<mode>", "vflls")
28242 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28243 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28244 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28245 ("*vec_cmphe<mode>_cc"): ... these.
28247 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28248 mode constant instead of magic value.
28250 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28252 * config/s390/s390.c (s390_expand_vec_compare): Support other
28253 vector floating point modes than just V2DF.
28254 (s390_expand_vcond): Likewise.
28255 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28256 (s390_cannot_change_mode_class): Prevent mode changes between TF
28257 and V1TF in vector registers.
28258 * config/s390/s390.md (DF, SF): New mode attributes.
28259 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28260 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28261 SFmode support for VRs.
28262 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28264 (VFT, VF_HW): New mode iterators.
28265 (vw, sdx): New mode attributes.
28266 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28267 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28268 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28269 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28270 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28271 also the new vector floating point modes. Renaming to ...
28273 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28274 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28275 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28276 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28277 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28278 ("vec_unordered<mode>"): ... these.
28280 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28281 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28282 ("*vec_extendv2df"): New insn definitions.
28284 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28286 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28287 ("mulditi3_2", "*muldi3_sign"): New patterns.
28288 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28289 rename the pattern definition.
28291 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28293 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28295 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28297 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28299 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28300 instruction if possible.
28301 * config/s390/vector.md (vec_halfnumelts): New mode
28303 ("*vec_vllezlf<mode>"): New pattern.
28305 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28307 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28308 ("popcountv4si2", "popcountv2di2"): Rename to ...
28309 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28310 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28312 ("popcount<mode>2_vxe"): New pattern.
28314 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28316 * common/config/s390/s390-common.c (processor_flags_table): Add
28318 * config.gcc: Add arch12.
28319 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28320 Default to arch12 for unknown CPU model numbers.
28321 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28322 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28323 PROCESSOR_max sanity check.
28324 * config/s390/s390-opts.h (enum processor_type): Add
28326 * config/s390/s390.c (processor_table): Add arch12.
28327 (s390_expand_builtin): Add check for B_VXE flag.
28328 (s390_issue_rate): Add PROCESSOR_ARCH12.
28329 (s390_get_sched_attrmask): Likewise.
28330 (s390_get_unit_mask): Likewise.
28331 (s390_sched_score): Enable z13 scheduling for arch12.
28332 (s390_sched_reorder): Likewise.
28333 (s390_sched_variable_issue): Likewise.
28334 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28336 (s390_tune_attr): Use z13 scheduling also for arch12.
28337 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28338 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28339 (TARGET_VXE_P): New macros.
28340 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28341 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28342 * config/s390/s390.opt: Add arch12 as processor_type.
28344 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28346 * config/s390/s390.md
28347 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28348 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28349 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28351 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28352 Rename expanders to ...
28354 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28355 ("fixuns_truncdddi2_emu"): ... these.
28357 ("fixuns_trunc<mode>si2_emu"): New expander.
28359 ("*fixuns_truncdfdi2_z13"): Rename to ...
28360 ("*fixuns_truncdfdi2_vx"): ... this.
28362 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28364 * config/s390/2964.md: Remove the single element vector compare
28365 instructions which are no longer used.
28366 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28368 (s390_canonicalize_comparison): Remove handling of DFmode
28370 (s390_expand_vec_compare_scalar): Remove function.
28371 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28372 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28374 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28376 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28378 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28380 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28381 will anyway by matched by mov<mode>_64dfp.
28383 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28385 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28386 vlef/vstef. Add missing operand to vleif.
28388 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28390 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28391 pair for all vector types with 64 bit elements.
28392 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28393 * config/s390/vector.md (V_HW_64): ... here.
28394 (V_128_NOSINGLE): New mode iterator.
28395 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28396 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28397 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28398 ("*vec_load_pairv2di"): Change to ...
28399 ("*vec_load_pair<mode>"): ... this one.
28401 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28403 * config/s390/constraints.md: Add comments.
28404 (jKK): Reject element sizes > 8 bytes.
28405 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28407 * config/s390/s390.md: Add the s_operand checks formerly in
28408 s390_split_ok_p to various splitters where they are still
28410 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28411 for 128 bit vectors. Plus two splitters.
28413 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28415 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28418 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28421 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28422 error if the boundary argument is not constant.
28424 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28426 PR rtl-optimization/80112
28427 * loop-doloop.c (doloop_condition_get): Don't check condition
28428 if cmp isn't SET with IF_THEN_ELSE src.
28430 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28432 PR tree-optimization/80158
28433 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28434 replacing a candidate statement, also replace it for the
28435 candidate's alternate interpretation.
28436 (replace_rhs_if_not_dup): Likewise.
28437 (replace_one_candidate): Likewise.
28439 2017-03-24 Richard Biener <rguenther@suse.de>
28441 PR tree-optimization/80167
28442 * graphite-isl-ast-to-gimple.c
28443 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28445 (translate_isl_ast_to_gimple::get_rename): Likewise.
28447 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28449 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28450 handling of certain combinations of target options, including the
28451 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28452 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28454 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28457 * config/arm/arm.md (*load_multiple): Add reload_completed to
28458 matching condition.
28460 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28461 Richard Biener <rguenth@suse.de>
28463 PR tree-optimization/79908
28464 PR tree-optimization/80136
28465 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28466 been cast away, gimplify_and_add suffices.
28468 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28470 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28472 2017-03-23 Richard Biener <rguenther@suse.de>
28474 PR tree-optimization/80032
28475 * gimplify.c (gimple_push_cleanup): Forced unconditional
28476 cleanups still have to go to the conditional_cleanups
28479 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28481 PR tree-optimization/80072
28482 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28484 (next_operand_entry_id): Change type to unsigned int.
28485 (sort_by_operand_rank): Make sure to return the right return value
28486 even if unsigned fields are bigger than INT_MAX.
28487 (struct oecount): Change cnt and id type to unsigned int.
28488 (oecount_hasher::equal): Formatting fix.
28489 (oecount_cmp): Make sure to return the right return value
28490 even if unsigned fields are bigger than INT_MAX.
28491 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28494 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28495 TREE_READONLY on result if writing it more than once.
28498 * doc/invoke.texi (-fsanitize=thread): Document that with
28499 -fnon-call-exceptions atomics are not able to throw
28503 * tsan.c: Include tree-eh.h.
28504 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28505 maybe_clean_or_replace_eh_stmt where needed.
28506 (instrument_memory_accesses): Add cfg_changed argument.
28507 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28508 if it returned true.
28509 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28511 PR rtl-optimization/63191
28512 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28513 wrapper function, moved the whole old content into ...
28514 (ix86_delegitimize_address_1): ... this. New inline function.
28515 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28516 true as last argument instead of ix86_delegitimize_address.
28518 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28520 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28521 cortexa57_branch_cost.
28523 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28525 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28527 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28530 * doc/md.texi (Constraints): Document wA constraint.
28531 * config/rs6000/constraints.md (wA): New.
28532 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28533 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28534 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28535 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28537 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28540 * gimplify.c (is_oacc_declared): New function.
28541 (oacc_default_clause): Use it to set default flags for acc declared
28542 variables inside parallel regions.
28543 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28544 declared variables.
28545 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28546 declare attribute to any decl as necessary.
28548 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28551 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28552 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28553 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28554 (arm_arch_lpae): This.
28555 * config/arm/arm.c (arm_arch7ve): Rename into ...
28556 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
28557 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
28560 2017-03-22 Martin Liska <mliska@suse.cz>
28563 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
28564 error message instead of an ICE.
28566 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28568 * doc/extend.texi (6.11 Additional Floating Types): Revise.
28570 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
28572 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28574 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28577 2017-03-21 Martin Sebor <msebor@redhat.com>
28579 * doc/extend.texi: Use "cannot" instead of "can't."
28580 * doc/hostconfig.texi: Same.
28581 * doc/install.texi: Same.
28582 * doc/invoke.texi: Same.
28583 * doc/loop.texi: Same.
28584 * doc/md.texi: Same.
28585 * doc/objc.texi: Same.
28586 * doc/rtl.texi: Same.
28587 * doc/tm.texi: Same.
28588 * doc/tm.texi.in: Same.
28589 * doc/trouble.texi: Same.
28591 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
28594 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
28595 (collect_checksum_attributes): Set it.
28596 (die_checksum_ordered): Use it.
28598 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28600 PR tree-optimization/79908
28601 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
28602 change: For a VA_ARG whose LHS has been cast away, use
28603 force_gimple_operand to construct the side effects.
28605 2017-03-21 David Malcolm <dmalcolm@redhat.com>
28607 PR translation/80001
28608 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
28609 more amenable to translation.
28610 (oacc_loop_auto_partitions): Likewise.
28612 2017-03-21 Marek Polacek <polacek@redhat.com>
28613 Martin Sebor <msebor@redhat.com>
28615 PR tree-optimization/80109
28616 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
28617 on INTEGRAL_TYPE_P.
28619 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28620 Segher Boessenkool <segher@kernel.crashing.org>
28623 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
28624 check reg_used_between_p between insn and one of succ or succ2
28625 depending on if succ is artificial insn not inserted into insn
28628 2017-03-21 Martin Liska <mliska@suse.cz>
28630 PR gcov-profile/80081
28631 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
28632 * doc/gcc.texi: Include gcov-dump stuff.
28633 * doc/gcov-dump.texi: New file.
28635 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
28637 PR rtl-optimization/79150
28638 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
28639 conditional jump, if the jump is the last insn of the loop.
28641 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28642 Richard Biener <rguenth@suse.de>
28644 PR tree-optimization/79908
28645 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28646 been cast away, use force_gimple_operand to construct the side
28649 2017-03-21 Martin Liska <mliska@suse.cz>
28651 PR libfortran/79956
28652 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
28655 2017-03-21 Brad Spengler <spender@grsecurity.net>
28658 * plugin.c (htab_hash_plugin): New function.
28659 (add_new_plugin): Use it and adjust.
28660 (parse_plugin_arg_opt): Adjust.
28661 (init_one_plugin): Likewise.
28663 2017-03-21 Richard Biener <rguenther@suse.de>
28665 PR tree-optimization/80032
28666 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
28667 if set force the cleanup to happen unconditionally.
28668 (gimplify_target_expr): Push inserted clobbers with force_uncond
28669 to avoid them being removed by control-dependent DCE.
28671 2017-03-21 Richard Biener <rguenther@suse.de>
28673 PR tree-optimization/80122
28674 * tree-inline.c (copy_bb): Do not expans va-arg packs or
28675 va_arg_pack_len when the inlined call stmt requires pack
28677 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
28679 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28682 * tsan.c (instrument_builtin_call): If the memory model argument
28683 is not a constant, assume it is valid.
28686 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
28689 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28691 PR rtl-optimization/79910
28692 * combine.c (can_combine_p): Do not allow combining an I0 or I1
28693 if its dest is used by an insn before I2 (other than the combined
28694 insns themselves, which are properly handled already).
28696 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28699 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28701 * combine.c (record_used_regs): New static function.
28702 (try_combine): Handle situations where there is an additional
28703 instruction between I2 and I3 which needs to have a LOG_LINK
28707 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28709 * combine.c (try_combine): Delete redundant i1 test. Call
28710 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28712 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28715 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
28716 alternatives 13/14.
28718 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28720 PR tree-optimization/80054
28721 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
28722 the optimization if a PHI or any of its arguments is not dominated
28723 by the candidate's basis. Use gphi* rather than gimple* as
28725 (replace_profitable_candidates): Clean up a gimple* variable that
28726 should be a gphi* variable.
28728 2017-03-20 Martin Sebor <msebor@redhat.com>
28731 * doc/extend.texi (attribute constructor): Document present limitation.
28733 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
28736 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
28737 __POWER9_VECTOR__ #ifdef control, change template definition to
28738 use Power9-specific built-in function.
28739 (vec_any_eq): Likewise.
28740 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28741 to control outcomes from this test.
28742 (vector_ae_<mode>p): For VEC_F modes, likewise.
28744 2017-03-20 Ian Lance Taylor <iant@google.com>
28746 * config/i386/i386.c (ix86_function_regparm): Save an extra
28747 register for -fsplit-stack with DECL_STATIC_CHAIN.
28749 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28752 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28753 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28755 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28757 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28759 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28762 2017-03-20 Marek Polacek <polacek@redhat.com>
28765 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28767 2017-03-20 Richard Biener <rguenther@suse.de>
28769 PR tree-optimization/80113
28770 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
28771 allocate extra SSA name for PHI def.
28772 (add_close_phis_to_outer_loops): Likewise.
28773 (add_close_phis_to_merge_points): Likewise.
28774 (copy_loop_close_phi_args): Likewise.
28775 (copy_cond_phi_nodes): Likewise.
28777 2017-03-20 Martin Liska <mliska@suse.cz>
28779 PR middle-end/79753
28780 * tree-chkp.c (chkp_build_returned_bound): Do not build
28781 returned bounds for a LHS that's not a BOUNDED_P type.
28783 2017-03-20 Martin Liska <mliska@suse.cz>
28787 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
28788 COMPLEX_CST and VECTOR_CST.
28790 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28793 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
28794 target operand. A new splitter adds the clobber statement in case
28795 the target operand is dead anyway.
28797 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
28799 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
28800 to age-old versions of binutils and glibc.
28802 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
28804 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
28806 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28808 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
28810 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28812 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
28813 requirement for binutils 2.13.
28815 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28817 * combine.c (try_combine): Delete redundant i1 test. Call
28818 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28820 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28822 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28823 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28825 <riscv64-*-elf>: Re-arrange section
28826 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28827 <riscv32-*-linux>: Likewise.
28828 <riscv64-*-elf>: Likewise
28829 <riscv64-*-linux>: Likewise.
28831 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28834 * aarch64.opt(verbose-cost-dump): Fix typo.
28836 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28839 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28840 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28842 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28844 * reload.c (find_reloads): When reloading a nonoffsettable address,
28845 use RELOAD_OTHER for it and its address reloads.
28847 PR rtl-optimization/79910
28848 * combine.c (record_used_regs): New static function.
28849 (try_combine): Handle situations where there is an additional
28850 instruction between I2 and I3 which needs to have a LOG_LINK
28853 2017-03-17 Jeff Law <law@redhat.com>
28855 PR tree-optimization/71437
28856 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28857 conditional in the hash table first.
28858 (vrp_dom_walker::before_dom_children): Extract condition from
28859 ASSERT_EXPR. Record condition, its inverion and any implied
28860 conditions as well.
28862 2017-03-17 Marek Polacek <polacek@redhat.com>
28863 Markus Trippelsdorf <markus@trippelsdorf.de>
28865 PR tree-optimization/80079
28866 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28869 2017-03-17 Richard Biener <rguenther@suse.de>
28871 PR middle-end/80075
28872 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28873 Properly verify the LHS before the RHS possibly claims to be
28875 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28878 2017-03-17 Martin Jambor <mjambor@suse.cz>
28880 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28881 (List of -O2 options): Likewise.
28882 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28885 2017-03-17 Tom de Vries <tom@codesourcery.com>
28887 * gcov-dump.c (print_usage): Print bug_report_url.
28889 2017-03-17 Richard Biener <rguenther@suse.de>
28891 PR middle-end/80050
28892 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28893 (parser::peek): Likewise.
28895 2017-03-17 Richard Biener <rguenther@suse.de>
28897 PR tree-optimization/80048
28898 * sese.c (free_sese_info): Properly release rename_map and
28899 copied_bb_map elements.
28901 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28903 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28904 Add linked-list forward and backlinks. Insert on
28905 construction, remove on destruction.
28906 (class pass_store_merging): Add m_stores_head field.
28907 (pass_store_merging::terminate_and_process_all_chains):
28908 Iterate over m_stores_head list.
28909 (pass_store_merging::terminate_all_aliasing_chains):
28911 (pass_store_merging::execute): Check for debug stmts first.
28912 Push new chains onto the m_stores_head stack.
28914 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28917 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28918 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28919 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28921 2017-03-16 Jeff Law <law@redhat.com>
28923 PR tree-optimization/71437
28924 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28925 member function. Implementation moved into after_dom_children
28926 member function and into the threader's thread_outgoing_edges
28928 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28929 some code into new thread_outgoing_edges.
28930 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28931 definition. Simplify marker handling (do it here). Assume we always
28932 have the available expression and the const/copies tables.
28933 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28935 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28936 * tree-vrp.c (equiv_stack): No longer file scoped.
28937 (vrp_dom_walker): New class.
28938 (vrp_dom_walker::before_dom_children): New member function.
28939 (vrp_dom_walker::after_dom_children): Likewise.
28940 (identify_jump_threads): Setup domwalker. Use it rather than
28941 walking edges in a random order by hand. Simplify setup/finalization.
28942 (finalize_jump_threads): Remove.
28943 (vrp_finalize): Do not call identify_jump_threads here.
28944 (execute_vrp): Do it here instead and call thread_through_all_blocks
28947 PR tree-optimization/71437
28948 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28950 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28952 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28953 (dom_opt_dom_walker::thread_across_edge): Remove
28954 handle_dominating_asserts argument. All callers changed.
28955 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28956 changes. Remove calls to lhs_of_dominating_assert. Other
28957 uses of handle_dominating_asserts turn into unconditional code
28958 (simplify_control_stmt_condition_1): Likewise.
28959 (simplify_control_stmt_condition): Likewise.
28960 (thread_through_normal_block, thread_across_edge): Likewise.
28961 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28962 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28963 object if it is not an SSA_NAME.
28964 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28965 before calling into the VRP specific simplifiers.
28966 (identify_jump_threads): Remove handle_dominating_asserts
28969 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28972 * tree-diagnostic.c (default_tree_printer): No longer static.
28973 * tree-diagnostic.h (default_tree_printer): New prototype.
28975 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28977 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28978 Change ins into fmov.
28980 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28982 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28983 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28984 Use h_con constraint for operand 1.
28985 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28986 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28988 2017-03-15 Jeff Law <law@redhat.com>
28990 PR tree-optimization/71437
28991 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28992 (record_temporary_equivalences): Use it.
28994 PR tree-optimization/71437
28995 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28996 tree-ssa-scopedtables.
28997 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28998 (record_conditions, record_cond, vuse_eq): Likewise.
28999 (record_edge_info): Adjust to API tweak of record_conditions.
29000 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29001 (record_temporary_equivalences, optimize_stmt): Likewise.
29002 (eliminate_redundant_computations): Likewise.
29003 (record_equivalences_from_stmt): Likewise.
29004 * tree-ssa-scopedtables.c: Include options.h and params.h.
29005 (vuse_eq): New function, moved from tree-ssa-dom.c
29006 (build_and_record_new_cond): Likewise.
29007 (record_conditions): Likewise. Accept vector of conditions rather
29008 than edge_equivalence structure for first argument.
29009 for the first argument.
29010 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29011 from tree-ssa-dom.c.
29012 (avail_exprs_stack::record_cond): Likewise.
29013 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29014 from tree-ssa-dom.c.
29015 (avail_exprs_stack): Add new member functions lookup_avail_expr
29017 (record_conditions): Declare.
29019 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29022 * lra-constraints.c (process_alt_operands): Increase reject for
29023 reloading an input/output operand.
29025 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29028 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29029 insns to convert from signed/unsigned char/short to IEEE 128-bit
29031 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29033 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29036 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29037 subreg of inner mode for values already in registers.
29039 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29041 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29042 iteration reg is used after the loop.
29044 2017-03-14 Martin Sebor <msebor@redhat.com>
29046 PR tree-optimization/79800
29047 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29048 precision in negative-positive range.
29049 (format_floating): Call non-const overload with adjusted precision.
29051 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29054 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29057 2017-03-14 Martin Sebor <msebor@redhat.com>
29059 PR middle-end/80020
29060 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29061 * builtins.def (aligned_alloc): Use it.
29064 * Makefile.in (GTFILES): Add calls.c.
29065 * calls.c: Include "gt-calls.h".
29067 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29069 PR rtl-optimization/79728
29070 * regs.h (struct target_regs): New field
29071 x_contains_allocatable_regs_of_mode.
29072 (contains_allocatable_regs_of_mode): New macro.
29073 * reginfo.c (init_reg_sets_1): Initialize it, and change
29074 contains_reg_of_mode so it includes global regs as well.
29075 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29076 rather than contains_regs_of_mode.
29078 2017-03-14 Martin Liska <mliska@suse.cz>
29080 * doc/invoke.texi: Document options that can't be combined with
29081 -fcheck-pointer-bounds.
29083 2017-03-14 Martin Liska <mliska@suse.cz>
29085 PR middle-end/79831
29086 * doc/invoke.texi (-Wchkp): Document the option.
29088 2017-03-14 Martin Liska <mliska@suse.cz>
29090 * Makefile.in: Install gcov-dump.
29092 2017-03-14 Martin Liska <mliska@suse.cz>
29094 * multiple_target.c (expand_target_clones): Bail out for
29095 an invalid attribute.
29097 2017-03-14 Richard Biener <rguenther@suse.de>
29099 * alias.c (struct alias_set_entry): Pack properly.
29100 * cfgloop.h (struct loop): Likewise.
29101 * cse.c (struct set): Likewise.
29102 * ipa-utils.c (struct searchc_env): Likewise.
29103 * loop-invariant.c (struct invariant): Likewise.
29104 * lra-remat.c (struct cand): Likewise.
29105 * recog.c (struct change_t): Likewise.
29106 * rtl.h (struct address_info): Likewise.
29107 * symbol-summary.h (function_summary): Likewise.
29108 * tree-loop-distribution.c (struct partition): Likewise.
29109 * tree-object-size.c (struct object_size_info): Likewise.
29110 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29111 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29112 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29113 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29114 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29115 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29116 (struct _stmt_vec_info): Likewise.
29118 2017-03-14 Martin Liska <mliska@suse.cz>
29121 * multiple_target.c (create_dispatcher_calls): Check that
29122 a target can create a function dispatcher.
29124 2017-03-14 Martin Liska <mliska@suse.cz>
29127 * multiple_target.c (expand_target_clones): Drop local.local
29128 flag for default implementation.
29130 2017-03-14 Richard Biener <rguenther@suse.de>
29132 PR tree-optimization/80030
29133 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29135 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29137 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29138 gcc_fallthrough() instead of __attribute__((fallthrough));
29140 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29142 * doc/gcc.texi: Remove "up" link to (DIR).
29143 * doc/gccint.texi: Ditto.
29145 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29147 * doc/install.texi (Specific) <avr>: Remove reference to
29150 2017-03-13 Jeff Law <law@redhat.com>
29152 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29153 attribute rather than comments.
29155 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29156 match_scratch operand is highest.
29158 2017-03-13 Martin Liska <mliska@suse.cz>
29160 PR middle-end/78339
29161 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29162 is a CHKP clone, use original declaration.
29164 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29166 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29167 (arc_conditional_register_usage): Use a different allocation order
29168 when optimizing for size.
29169 * common/config/arc/arc-common.c (arc_option_optimization_table):
29170 Section anchors default on when optimizing for size.
29172 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29174 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29176 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29178 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29179 * config/arc/arc.md (cpu_facility): Add cd variant.
29180 (*movqi_insn): Add code density variant.
29181 (*movhi_insn): Likewise.
29182 (*movqi_insn): Likewise.
29183 (*addsi3_mixed): Likewise.
29184 (subsi3_insn): Likewise.
29186 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29188 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29190 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29192 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29193 expressions with MINUS and UNARY ops.
29195 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29198 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29200 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29201 between vec_select and vector argument.
29202 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29203 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29204 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29205 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29206 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29207 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29209 2017-03-13 Richard Biener <rguenther@suse.de>
29212 * params.def (vect-max-peeling-for-alignment): Fix typo.
29214 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29216 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29217 issue that only occurred with binutils below 2.18.
29219 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29221 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29222 refer to binutils 2.11/2.12 minimum.
29224 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29226 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29227 ftp.kernel.org and simplify binutils requirement.
29229 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29231 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29233 (Optimize Options): Ditto. Also remove redundancy.
29235 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29237 PR translation/79848
29238 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29240 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29241 to G_ to avoid double translation.
29243 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29245 PR translation/79923
29246 * auto-profile.c (get_combined_location): Convert leading
29247 character of diagnostics to lower case and remove trailing period.
29248 (read_profile): Likewise for various diagnostics.
29249 * config/arm/arm.c (arm_option_override): Remove trailing period
29250 from various diagnostics.
29251 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29252 (msp430_expand_delay_cycles): Likewise.
29254 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29257 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29258 full command-line argument, rather than just "str".
29259 (aarch64_validate_march): Likewise.
29260 (aarch64_validate_mtune): Likewise.
29262 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29264 PR rtl-optimization/78911
29265 * lra-assigns.c (must_not_spill_p): New function.
29266 (spill_for): Use it.
29268 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29270 PR tree-optimization/79981
29271 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29272 ATOMIC_COMPARE_EXCHANGE ifn result.
29273 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29274 IFN_ATOMIC_COMPARE_EXCHANGE.
29276 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29279 * opts.c (parse_sanitizer_options): Add missing question mark to
29280 "did you mean" message.
29282 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29284 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29286 (VMULEUH_UNS): Likewise.
29287 (VMULOUB_UNS): Likewise.
29288 (VMULOUH_UNS): Likewise.
29289 * config/rs6000/rs6000.c (builtin_function_type): Remove
29290 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29292 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29295 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29296 x with result of extra_parsing_for_operand_code_0.
29297 (function_reader::extra_parsing_for_operand_code_0): Convert
29298 return type from void to rtx, returning x. When reading
29299 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29300 larger size containing struct block_symbol.
29302 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29304 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29305 -mfloat128-hardware without -m64.
29307 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29310 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29311 entries to the case statement that marks unsigned arguments to
29312 overloaded functions.
29314 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29316 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29317 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29319 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29322 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29323 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29325 2017-03-10 Martin Liska <mliska@suse.cz>
29329 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29330 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29333 2017-03-10 Olivier Hainque <hainque@adacore.com>
29335 * tree-switch-conversion (array_value_type): Start by resetting
29336 candidate type to it's main variant.
29338 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29340 PR rtl-optimization/79909
29341 * combine.c (try_combine): Use simplify_replace_rtx on individual
29342 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29343 of the whole CALL_INSN_FUNCTION_USAGE.
29345 PR tree-optimization/79972
29346 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29347 get_range_info on SSA_NAMEs. Formatting fixes.
29349 2017-03-10 Richard Biener <rguenther@suse.de>
29350 Jakub Jelinek <jakub@redhat.com>
29352 PR tree-optimization/77975
29353 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29354 edge to be constant.
29355 (get_val_for): For constant x return it. Formatting fix.
29356 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29357 would use the same bases as the current one.
29359 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29361 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29362 instead of vec_select for V1TImode.
29363 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29365 (VSX_LE_128): Add V1TI to this mode iterator.
29366 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29367 (*vsx_le_perm_store_<mode>): Likewise.
29368 (pre-reload splitter for VSX stores): Likewise.
29369 (post-reload splitter for VSX stores): Likewise.
29370 (*vsx_xxpermdi2_le_<mode>): Likewise.
29371 (*vsx_lxvd2x2_le_<mode>): Likewise.
29372 (*vsx_stxvd2x2_le_<mode>): Likewise.
29374 2017-03-09 Michael Eager <eager@eagercon.com>
29376 Correct failures with --enable-checking=yes,rtl.
29378 * config/microblaze/microblaze.c (microblaze_expand_shift):
29379 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29380 test for const0_rtx.
29381 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29382 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29384 2017-03-09 Richard Biener <rguenther@suse.de>
29386 PR tree-optimization/79977
29387 * graphite-scop-detection.c (scop_detection::merge_sese):
29388 Handle the case of extra exits to blocks dominating the entry.
29390 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29392 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29395 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29397 PR rtl-optimization/79949
29398 * lra-constraints.c (process_alt_operands): Check memory when
29399 trying to predict a cycle. Print about the overall increase.
29401 2017-03-09 Richard Biener <rguenther@suse.de>
29403 PR middle-end/79971
29404 * gimple-expr.c (useless_type_conversion_p): Preserve
29405 TYPE_SATURATING for fixed-point types.
29407 2017-03-09 Richard Biener <rguenther@suse.de>
29410 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29411 alignment of BLKmode params.
29413 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29416 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29417 (VALL_NO_V2Q): Likewise.
29419 * config/aarch64/aarch64-simd.md
29420 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29422 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29423 VALL_NO_V2Q mode iterator.
29424 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29426 2017-03-09 Martin Liska <mliska@suse.cz>
29428 PR tree-optimization/79631
29429 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29430 tree_int_cst_sign_bit just for INTEGER constants.
29432 2017-03-09 Martin Liska <mliska@suse.cz>
29436 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29439 2017-03-09 Marek Polacek <polacek@redhat.com>
29442 * tree.c (inchash::add_expr): Handle TREE_VEC.
29444 2017-03-09 Martin Liska <mliska@suse.cz>
29447 (chkp_narrow_size_and_offset): New function.
29448 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29449 (void chkp_parse_bit_field_ref): New function.
29450 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29451 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29453 2017-03-09 Martin Liska <mliska@suse.cz>
29456 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29457 (chkp_find_bounds_1): Remove gcc_unreachable.
29459 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29462 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29463 BUILT_IN_SYNC*, determine the access type from the size suffix and
29464 always build a MEM_REF with that type. Handle forgotten
29465 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29468 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29469 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29470 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29471 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29472 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29473 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29474 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29475 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29476 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29477 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29478 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29479 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29480 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29481 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29482 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29483 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29484 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29485 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29486 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29487 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29488 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29489 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29490 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29491 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29492 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29493 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29494 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29495 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29496 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29497 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29498 definitions outside of __OPTIMIZE__ guarded section.
29501 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29502 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29503 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29504 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29507 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29509 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29510 ("vfenez<mode>"): Add missing constraints.
29512 2017-03-08 Martin Sebor <msebor@redhat.com>
29515 * config/nds32/nds32.c (nds32_option_override):
29516 Fix misspelled diagnostic.
29518 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29521 * gimplify.c (gimplify_omp_for): Replace index var in outer
29522 taskloop statement with an artificial variable and add
29523 OMP_CLAUSE_PRIVATE clause for it.
29525 2017-03-08 Richard Biener <rguenther@suse.de>
29527 PR tree-optimization/79955
29528 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29529 for accesses that are completely outside of the variable.
29531 2017-03-08 Andrew Haley <aph@redhat.com>
29533 PR tree-optimization/79943
29534 * tree-ssa-loop-split.c (compute_new_first_bound): When
29535 calculating the new upper bound, (END-BEG) should be added, not
29538 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29540 * config/avr/avr.md (setmemhi): Make sure match_dup
29541 operand number comes before match_scratch.
29543 2017-03-08 Richard Biener <rguenther@suse.de>
29545 PR tree-optimization/79920
29546 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29547 with ncopies == 1 to ...
29548 (vect_transform_slp_perm_load): ... here. Properly compute
29549 all element loads by iterating VF times over the group. Do
29550 not handle ncopies (computed in a broken way) in
29551 vect_create_mask_and_perm.
29553 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29556 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
29557 is a uniform vector, use uniform_vector_p return value instead of
29558 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
29560 2017-03-07 Marek Polacek <polacek@redhat.com>
29562 PR middle-end/79809
29563 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
29564 (alloca_call_type): Likewise.
29566 2017-03-07 Martin Liska <mliska@suse.cz>
29568 * gcov.c (process_args): Put comment to correct location.
29570 2017-03-07 Martin Liska <mliska@suse.cz>
29572 PR middle-end/68270
29573 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
29574 Use array_at_struct_end_p instead of DECL_CHAIN (field).
29575 (chkp_narrow_bounds_for_field): Likewise.
29576 (chkp_parse_array_and_component_ref): Pass one more argument to
29579 2017-03-07 Richard Biener <rguenther@suse.de>
29581 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
29584 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
29586 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
29587 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
29589 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29592 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
29593 to end of description.
29594 (PARAM_MAX_STORES_TO_MERGE): Likewise.
29596 2017-03-07 Jakub Jelinek <jakub@redhat.com>
29598 PR rtl-optimization/79901
29599 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
29601 (*avx512f_<code><mode>3<mask_name>): ... this.
29602 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
29603 iterator instead of VI8_AVX2_AVX512BW.
29605 PR rtl-optimization/79901
29606 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
29607 min/max expander, expand it using expand_vec_cond_expr.
29610 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
29613 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29616 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
29617 to void * for PCH reasons.
29618 * dwarf2out.c (output_loc_operands, output_die): Cast
29619 v.val_vec.array to unsigned char *.
29621 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
29624 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
29627 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
29629 PR rtl-optimization/79571
29630 * lra-constraints.c (process_alt_operands): Calculate static
29631 reject and subtract it from overall when only addresses will be
29634 2017-03-06 Julia Koval <julia.koval@intel.com>
29637 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
29638 incoming stack boundary to 128 for 64-bit targets.
29640 2017-03-06 Richard Biener <rguenther@suse.de>
29642 PR tree-optimization/79894
29643 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
29644 to NULL after folding it.
29646 2017-03-06 Richard Biener <rguenther@suse.de>
29648 PR tree-optimization/79824
29649 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
29650 check disabling peeling for gaps.
29652 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
29654 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
29655 attributes): Document gettimeofday.
29657 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29659 * config/s390/s390.c (s390_option_override_internal): Set
29660 PARAM_MIN_VECT_LOOP_BOUND
29662 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29664 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
29665 * config/s390/s390.md: Likewise.
29667 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29670 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
29671 (<avx2_avx512>_perm<mode>): Rename to ...
29672 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
29674 (<avx512>_perm<mode>_mask): Rename to ...
29675 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
29677 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
29678 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
29679 instead of VI8F_256_512.
29680 (avx512f_perm<mode>): New define_expand.
29681 (avx512f_perm<mode>_mask): Likewise.
29682 (avx512f_perm<mode>_1<mask_name>): New define_insn.
29683 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
29685 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29687 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
29688 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
29690 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
29692 2017-03-06 Martin Liska <mliska@suse.cz>
29695 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
29696 when having a SSA NAME w/o VAR_DECL assigned to it.
29698 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29700 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
29701 msa_dpsub_<su>_d): Fix MODE for vec_select.
29703 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29705 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
29707 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
29709 2017-03-06 Richard Biener <rguenther@suse.de>
29711 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
29712 * plugin.c (register_plugin_info): Likewise.
29713 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
29715 2017-03-05 Jakub Jelinek <jakub@redhat.com>
29717 * config/i386/sse.md (sse_storehps, sse_storelps,
29718 avx_<castmode><avxsizesuffix>_<castmode>,
29719 avx512f_<castmode><avxsizesuffix>_<castmode>,
29720 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
29721 in condition that at least one operand is not a MEM.
29723 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29725 PR middle-end/79805
29726 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
29727 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
29729 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
29730 gimple_call_nothrow_p flag based on whether original builtin can throw.
29731 If it can, emit following stmts on the fallthrough edge.
29732 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
29733 don't create new bb if inserting just debug stmts on the edge, try to
29734 insert them on the fallthru bb or just reset debug stmts.
29736 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
29739 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29740 restore recog_data (including the operand rtxes inside it) around
29741 the call to get_insn_template.
29743 2017-03-03 Martin Sebor <msebor@redhat.com>
29745 PR tree-optimization/79699
29746 * context.c (context::~context): Free MPFR caches to avoid
29747 a memory leak on program exit.
29749 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29751 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29752 Use wide_int::ulow () instead of .elt (0).
29754 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29756 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29757 (*pushxf): Limit oF constraint to 32bit targets and add oC
29758 constraint for 64bit targets.
29759 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29760 (*pushdf): Change rmF constraint to rmC.
29762 2017-03-03 Martin Liska <mliska@suse.cz>
29764 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29765 Remove unused variable.
29767 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29770 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
29771 is a memory operand, increase num_memory.
29772 (ix86_expand_args_builtin): Likewise.
29774 2017-03-03 Jan Hubicka <jh@suse.cz>
29777 * ipa-devirt.c (maybe_record_node): Properly handle
29778 __cxa_pure_virtual visibility.
29780 2017-03-03 Martin Liska <mliska@suse.cz>
29782 PR tree-optimization/79803
29783 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
29785 (pass_loop_prefetch::execute): Disabled optimization if an
29786 assumption about L1 cache size is not met.
29788 2017-03-03 Martin Liska <mliska@suse.cz>
29790 PR rtl-optimization/79574
29791 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
29792 (hash_scan_set): Likewise.
29793 (dump_hash_table): Likewise.
29794 (hoist_code): Likewise.
29796 2017-03-03 Richard Biener <rguenther@suse.de>
29798 * fixed-value.c (fixed_from_string): Restore use of elt (1)
29799 in place of uhigh ().
29800 (fixed_convert_from_real): Likewise.
29802 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29805 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
29807 2017-03-03 Richard Biener <rguenther@suse.de>
29809 PR middle-end/79818
29810 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
29811 TYPE_OVERFLOW_UNDEFINED check.
29813 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29815 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
29817 (vector_ae_<mode>_p): Likewise.
29818 (vector_nez_<mode>_p): Likewise.
29819 (vector_ne_v2di_p): Likewise.
29820 (vector_ae_v2di_p): Likewise.
29821 (vector_ne_<mode>_p): Likewise.
29822 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29824 (vsx_tsqrt<mode>2_fe): Likewise.
29826 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29829 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29831 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29833 PR rtl-optimization/79780
29834 * cprop.c (one_cprop_pass): When second and further conditional trap
29835 in a single basic block is turned into an unconditional trap, turn it
29836 into a deleted note to avoid RTL verification failures.
29838 2017-03-02 Richard Biener <rguenther@suse.de>
29840 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29842 2017-03-02 Richard Biener <rguenther@suse.de>
29844 PR tree-optimization/79345
29846 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29847 param and abort the walk, returning -1 if it is hit.
29848 (walk_aliased_vdefs): Take a limit param and pass it on.
29849 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29850 defaulting to 0 and return a signed int.
29851 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29852 (check_defs): New helper.
29853 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29854 about uninitialized memory.
29855 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29856 bogus uninitialized warning.
29857 (fixed_convert_from_real): Likewise.
29859 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29861 PR tree-optimization/66768
29862 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29863 iv_use if base object can't be determined.
29865 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29867 PR tree-optimization/79345
29868 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29869 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29870 (get_pattern_stats): Initialize it.
29871 * genemit.c (gen_expand): Verify match_scratch numbers come after
29872 match_operand/match_dup numbers.
29873 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29874 match_scratch numbers.
29875 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29877 * config/s390/s390.md (trunctdsd2): Likewise.
29879 2017-03-02 Richard Biener <rguenther@suse.de>
29881 * wide-int.h (wide_int_storage::operator=): Implement in terms
29884 2017-03-02 Richard Biener <rguenther@suse.de>
29886 PR tree-optimization/79777
29887 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29888 the to insert expression to sth existing.
29890 2017-03-01 Martin Sebor <msebor@redhat.com>
29892 PR middle-end/79692
29893 * gimple-ssa-sprintf.c
29894 (directive::known_width_and_precision): New function.
29895 (format_integer): Use it.
29896 (get_mpfr_format_length): Consider the full range of precision
29897 when computing %g output with the # flag. Set the likely byte
29898 count to 3 rather than 1 when precision is indeterminate.
29899 (format_floating): Correct the lower bound of precision.
29901 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29903 * doc/invoke.texi: Document default code model for 64-bit Linux.
29905 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29908 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29909 udiv rather than div since input pattern is unsigned.
29911 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29913 * config/i386/i386.c (print_reg): Warn for values of
29914 unsupported size in integer register.
29916 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29919 * config/rs6000/predicates.md (current_file_function_operand): Do
29920 not allow self calls to be local if the function is replaceable.
29922 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29925 * config/rs6000/altivec.h (vec_ctz and others): Change the
29926 preprocessor macro that controls conditional compilation from
29927 _ARCH_PWR9 to __POWER9_VECTOR__.
29928 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29929 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29930 control (instead of _ARCH_PWR9 control) so that template
29931 definition uses power9-specific function.
29932 (vec_any_eq): Likewise.
29933 (vec_all_ne): Change macro definition to use a power9-specific
29934 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29935 _ARCH_PWR9 control).
29936 (vec_any_eq) Likewise.
29937 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29938 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29939 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29940 support for xvcmpnedp instruction.
29941 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29942 macro expansion so that Power9 implementation of vec_all_ne does
29943 not use the AltiVec predicate framework.
29944 (VCMPNEH_P): Likewise.
29945 (VCMPNEW_P): Likewise.
29946 (VCMPNED_P): Likewise.
29947 (VCMPNEFP_P): Likewise.
29948 (VCMPNEDP_P): Likewise.
29949 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29950 implementation of vec_any_eq to not use AltiVec predicate
29952 (VCMPAEH_P): Likewise.
29953 (VCMPAEW_P): Likewise.
29954 (VCMPAED_P): Likewise.
29955 (VCMPAEFP_P): Likewise.
29956 (VCMPAEDP_P): Likewise.
29957 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29958 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29959 not use the AltiVec predicate framework.
29960 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29961 of vec_any_eq to not use AltiVec predicate framework.
29962 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29963 support for predefined __POWER9_VECTOR__ macro to indicate that
29964 Power9 instruction selection is enabled.
29965 (altivec_overloaded_builtins): Remove extraneous
29966 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29967 function argument types RS6000_BTI_bool_V16QI and
29968 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29969 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29970 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29971 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29972 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29973 Power9 for implementations of vec_cmpne. Change the signature for
29974 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29975 (representing vec_all_ne) to remove the previously described first
29976 argument of type RS6000_BTI_INTSI, as this was an artifact of
29977 reliance on the AltiVec predicate framework, which is no longer
29978 used in the implementation of these functions. Add
29979 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29980 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29981 since, unlike the AltiVec predicate framework implementation, we
29982 do not share function descriptors between vec_alle and vec_anyeq.
29983 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29984 set of modes that receive special treatment even when
29985 TARGET_P9_VECTOR is true. The special treatment emits code that
29986 does not depend on Power9 instructions.
29987 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29988 define_expand to not rely on AltiVec predicate framework.
29989 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29991 (vector_ne_v2di_p): Change this define_expand to not rely on
29992 AltiVec predicate framework.
29993 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29995 (vector_ne_<mode>_p): Change this define_expand to not rely on
29996 AltiVec predicate framework.
29997 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29999 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30000 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30001 define_insn pattern.
30002 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30003 define_insn pattern because the xvcmpne<VSs>. instruction is not
30005 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30006 instruction is not supported.
30008 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30010 * config/nvptx/nvptx.c: Include intl.h.
30012 2017-03-01 Martin Jambor <mjambor@suse.cz>
30015 * ipa-prop.h (ipa_bits): Removed field known.
30016 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30017 to pointers. Adjusted their comments to warn about their sharing.
30018 (ipcp_transformation_summary): Change bits to a vector of pointers.
30019 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30020 (ipa_get_ipa_bits_for_value): Declare.
30021 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30022 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30023 (ipa_bits_hash_table): Likewise.
30024 (ipa_vr_ggc_hash_traits): Likewise.
30025 (ipa_vr_hash_table): Likewise.
30026 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30027 being pointers and vr_known being removed.
30028 (ipa_set_jf_unknown): Likewise.
30029 (ipa_get_ipa_bits_for_value): New function.
30030 (ipa_set_jfunc_bits): Likewise.
30031 (ipa_get_value_range): New overloaded functions.
30032 (ipa_set_jfunc_vr): Likewise.
30033 (ipa_compute_jump_functions_for_edge): Use the above functions to
30034 construct bits and vr parts of jump functions.
30035 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30036 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30038 (ipcp_grow_transformations_if_necessary): Also allocate
30039 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30041 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30042 them. Fix too long lines.
30043 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30044 vr_known being removed.
30045 (ipa_read_jump_function): Use new setter functions to construct bits
30046 and vr parts of jump functions or set them to NULL.
30047 (write_ipcp_transformation_info): Adjust for bits being pointers.
30048 (read_ipcp_transformation_info): Likewise.
30049 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30051 Include gt-ipa-prop.h.
30052 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30054 (ipcp_store_bits_results): Likewise.
30055 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30056 Do not write to existing jump functions but use a temporary instead.
30058 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30061 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30062 attempt to use its first operand as BIT_FIELD_REF base.
30064 2017-03-01 Richard Biener <rguenther@suse.de>
30066 PR middle-end/79721
30067 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30068 interpolating formula in wrapping arithmetic.
30069 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30071 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30073 PR tree-optimization/79734
30074 * tree-vect-generic.c (expand_vector_condition): Optimize
30075 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30076 Handle VEC_COND_EXPR where comparison has different inner width from
30077 type's inner width.
30079 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30081 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30082 markup, and similar issues. Remove @opindex entries for things
30083 that aren't options. Add missing -mmpy-option entries.
30085 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30087 PR tree-optimization/79737
30088 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30089 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30090 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30091 instead of byte_size. Formatting fix.
30092 (shift_bytes_in_array_right): Formatting fix.
30094 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30097 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30098 condition on optimize for the leaf function test.
30100 2017-02-28 Martin Liska <mliska@suse.cz>
30103 * read-rtl-function.c (function_reader::handle_unknown_directive):
30104 Bail out when one uses -flto.
30106 2017-02-28 Martin Liska <mliska@suse.cz>
30108 * common.opt: Replace space with tabular for options of <number>
30110 * config/i386/i386.opt: Show <number> value for
30111 -mlarge-data-threshold.
30112 * opts.c (print_filtered_help): Do not display number in hexadecimal
30115 2017-02-28 Martin Liska <mliska@suse.cz>
30117 * common.opt: Fix --help=option -Q for options which are of
30120 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30122 * config/i386/i386.c (print_reg): Error out for values
30123 of 8-bit size in invalid integer register.
30125 2017-02-28 Martin Sebor <msebor@redhat.com>
30127 PR tree-optimization/79691
30128 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30130 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30133 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30134 gcc_unreachable with output_operand_lossage.
30136 2017-02-28 Richard Biener <rguenther@suse.de>
30138 PR tree-optimization/79740
30139 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30141 (visit_nary_op): Insert the nary into the hashtable if we
30142 pattern-matched sth.
30143 * tree-ssa-pre.c (eliminate_insert): Robustify.
30145 2017-02-28 Richard Biener <rguenther@suse.de>
30147 PR middle-end/79731
30148 * fold-const.c (decode_field_reference): Reject out-of-bound
30151 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30153 * config/i386/i386.c: Include intl.h.
30154 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30155 instead of just cond ? "..." : "...".
30156 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30157 * coverage.c (read_counts_file): Likewise.
30158 * omp-offload.c: Include intl.h.
30159 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30160 of just cond ? "..." : "...".
30161 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30162 of just cond ? "..." : "...".
30164 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30167 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30169 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30170 'tune for' CPU name.
30171 * config/arm/arm-cpu-data.h: Regenerated.
30173 2017-02-28 Richard Biener <rguenther@suse.de>
30175 PR tree-optimization/79732
30176 * tree-inline.c (expand_call_inline): Do not shadow var.
30178 2017-02-28 Richard Biener <rguenther@suse.de>
30180 PR tree-optimization/79723
30181 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30182 address-space properly.
30184 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30186 * doc/optinfo.texi (Optimization groups): Fix option used for
30188 * doc/invoke.texi (-fopt-info): Document "omp".
30189 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30190 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30191 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30192 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30193 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30195 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30197 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30198 OPTGROUP_OMP with "omp".
30200 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30203 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30204 for arithmetic shift of unsigned V2DI.
30206 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30208 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30209 arc/linux.h headers.
30210 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30211 (LINK_SPEC): Likewise.
30212 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30213 (EXTRA_SPECS): Likewise.
30214 (STARTFILE_SPEC): Likewise.
30215 (ENDFILE_SPEC): Likewise.
30216 (LIB_SPEC): Likewise.
30217 (TARGET_SDATA_DEFAULT): Likewise.
30218 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30219 (MULTILIB_DEFAULTS): Likewise.
30220 (DWARF2_UNWIND_INFO): Likewise.
30221 * config/arc/big.h: New file.
30222 * config/arc/elf.h: Likewise.
30223 * config/arc/linux.h: Likewise.
30224 * config/arc/t-uClibc: Remove.
30226 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30228 PR tree-optimization/77536
30229 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30230 (tree_transform_and_unroll_loop): Use above function to compute the
30231 estimated niter of unrolled loop and use it when scaling profile.
30232 Also use count info rather than frequency if it's non-zero.
30233 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30234 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30235 (vect_transform_loop): Call above function.
30237 2017-02-27 Richard Biener <rguenther@suse.de>
30239 PR tree-optimization/45397
30240 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30241 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30242 (visit_nary_op): Add pattern matching for CSEing sign-changed
30243 or truncated operations with wider ones.
30245 2017-02-27 Richard Biener <rguenther@suse.de>
30247 PR tree-optimization/79690
30248 * tree-vect-stmts.c (vectorizable_store): Use vector type
30249 built from the DR with address-space.
30251 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30253 * doc/invoke.texi (Optimize Options): Refine the description
30254 of asan-use-after-return.
30256 2017-02-25 Alan Modra <amodra@gmail.com>
30258 PR rtl-optimization/79584
30259 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30260 base, not ad->base_term, the reg within base. Remove assertion
30261 that ad->base == ad->base_term. Replace gen_int_mode using
30262 bogus mode with const0_rtx.
30264 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30266 PR middle-end/79396
30267 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30268 FMA_EXPR like tcc_binary or tcc_unary.
30270 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30273 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30275 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30276 (output_loc_operands): Handle DW_OP_call_ref and
30277 DW_OP_GNU_variable_value.
30278 (struct variable_value_struct): New type.
30279 (struct variable_value_hasher): Likewise.
30280 (variable_value_hash): New variable.
30281 (string_types): Remove.
30282 (copy_loc_descr): New function.
30283 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30284 (prepend_loc_descr_to_each): New function.
30285 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30286 instead of add_loc_descr_to_each if the first argument is single
30287 location list and the second has multiple.
30288 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30289 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30290 when looking for variable value which doesn't have other location info.
30291 (loc_list_from_tree): Formatting fix.
30292 (gen_array_type_die): Simplify DW_AT_string_length handling.
30293 (adjust_string_types): Remove.
30294 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30295 string_types. Call resolve_variable_values.
30296 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30297 (resolve_addr_in_expr): Likewise. Add A argument.
30298 (copy_deref_exprloc): Remove deref argument. Adjust for the
30299 original expression being DW_OP_GNU_variable_value with optionally
30300 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30301 optionally after it.
30302 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30303 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30304 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30305 (variable_value_hasher::hash, variable_value_hasher::equal): New
30307 (resolve_variable_value_in_expr, resolve_variable_value,
30308 resolve_variable_values, note_variable_value_in_expr,
30309 note_variable_value): New functions.
30310 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30313 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30316 * opts.h (handle_generated_option): Add GENERATED_P argument.
30317 * opts-common.c (handle_option): Adjust function comment.
30318 (handle_generated_option): Add GENERATED_P argument, pass it to
30320 (control_warning_option): Pass false to handle_generated_option
30322 * opts.c (maybe_default_option): Pass true to handle_generated_option
30324 * optc-gen.awk: Likewise.
30326 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30328 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30329 a REG, look at the REG it is a SUBREG of.
30330 (splitter for cmpeqsi_t): Ditto.
30332 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30334 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30335 the special USEs with the pattern of the insn, not the insn itself.
30337 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30340 * doc/invoke.texi: Document -mload-store-pairs.
30342 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30343 Sandra Loosemore <sandra@codesourcery.com>
30345 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30346 argument isn't a CONST_INT.
30347 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30348 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30349 (nios2_validate_compare): Bypass alternate compare logic if *op2
30350 is not a CONST_INT.
30351 (ldstwm_operation_p): Return false if first_base is not a REG or
30352 if first_offset is not a CONST_INT.
30354 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30356 * config/cris/cris.md: Use correct operand in a define_peephole2.
30358 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30360 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30362 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30364 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30365 this_insn if it is an INSN or JUMP_INSN.
30366 (force_offsettable): Look at base, not at addr.
30367 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30368 on things that aren't necessarily CONST_INTs.
30370 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30372 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30373 -mfpmath=sse is the default also for x86-32 targets with SSE2
30374 instruction set when @option{-ffast-math} is enabled
30376 2017-02-24 Jeff Law <law@redhat.com>
30378 PR rtl-optimizatoin/79286
30379 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30382 2017-02-24 Richard Biener <rguenther@suse.de>
30384 PR tree-optimization/79389
30385 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30388 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30390 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30391 function comment to reflect reality.
30392 (loop_exits_before_overflow): Fix typo in function description.
30394 2017-02-24 Richard Biener <rguenther@suse.de>
30396 PR tree-optimization/79389
30397 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30398 properly that a threading opportunity exists. Detect conditional
30399 copy/constant propagation opportunities.
30401 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30403 * config/visium/visium.md (type): Add trap.
30404 (b): New mode attribute.
30405 (*btst): Rename into...
30406 (*btst<mode>): ...this and adjust.
30407 (*cbranchsi4_btst_insn): Rename into...
30408 (*cbranch<mode>4_btst_insn): ...this and adjust.
30409 (trap): New define_insn.
30411 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30413 PR tree-optimization/79389
30414 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30415 (noce_reversed_cond_code): New function.
30416 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30417 reversed_comparison_code. Formatting fix.
30418 (noce_try_store_flag): Test rev_cond != NULL in addition to
30419 reversed_comparison_code.
30420 (noce_try_store_flag_constants): Likewise.
30421 (noce_try_store_flag_mask): Likewise.
30422 (noce_try_addcc): Use rev_cond if non-NULL instead of
30423 reversed_comparison_code.
30424 (noce_try_cmove_arith): Likewise. Formatting fixes.
30425 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30426 (noce_find_if_block): Initialize rev_cond.
30427 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30428 instead of false as last argument never attempt to reverse it
30431 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30433 PR tree-optimization/79663
30434 * tree-predcom.c (combine_chains): Process refs in reverse order
30435 only for ZERO length chains, and add explaining comment.
30437 2017-02-23 Jeff Law <law@redhat.com>
30439 PR tree-optimization/79578
30440 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30441 in call to operand_equal_p.
30443 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30446 * config/i386/cpuid.h: Fix another undefined behavior.
30448 2017-02-23 Richard Biener <rguenther@suse.de>
30450 PR tree-optimization/79683
30451 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30452 vector types for data-refs.
30454 2017-02-23 Martin Liska <mliska@suse.cz>
30456 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30458 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30460 PR middle-end/79665
30461 * internal-fn.c (get_range_pos_neg): Moved to ...
30462 * tree.c (get_range_pos_neg): ... here. No longer static.
30463 * tree.h (get_range_pos_neg): New prototype.
30464 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30465 are known to be in between 0 and signed maximum inclusive, try to
30466 expand both unsigned and signed divmod and use the cheaper one from
30469 2017-02-22 Jeff Law <law@redhat.com>
30471 PR tree-optimization/79578
30472 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30473 to compare base operands.
30475 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30478 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30479 gpc_reg_operand instead of fpr_reg_operand.
30481 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30483 * config/mips/mips.c (mips_return_in_memory): Force FP
30484 vector types to be returned in memory for o32 ABI.
30486 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30488 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30489 instead of DW_TAG_member for static data member declarations and don't
30490 set no_linkage_name for static inline data members.
30491 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30494 2017-02-22 Martin Liska <mliska@suse.cz>
30496 * doc/invoke.texi: Replace inequality signs with square brackets
30499 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30502 * lra-constraints.c (simplify_operand_subreg): Handle
30503 WORD_REGISTER_OPERATIONS targets.
30505 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30508 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30509 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30510 elimination by swapping fld*.
30512 2017-02-22 Richard Biener <rguenther@suse.de>
30514 PR tree-optimization/79673
30515 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30516 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30517 irrelevant address-space qualifiers and avoiding a
30518 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30520 2017-02-22 Richard Biener <rguenther@suse.de>
30522 PR tree-optimization/79666
30523 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30524 to not symbolically negate if that may introduce undefined
30527 2017-02-22 Martin Liska <mliska@suse.cz>
30530 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30531 * data-streamer-out.c (streamer_write_gcov_count_stream):
30533 * value-prof.c (stream_out_histogram_value): Make assert more
30534 precise based on type of counter.
30536 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30539 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30540 Use nonimmediate_operand instead of memory_operand for operand 1.
30541 (float-extend standard_x87sse_constant_load splitter): Ditto.
30543 2017-02-21 Jeff Law <law@redhat.com>
30545 PR tree-optimization/79621
30546 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30547 blocks with edges to themselves.
30549 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30552 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30553 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30554 Use gimple_call_builtin_p.
30557 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
30558 on temporarily removed DEBUG_INSNs.
30560 PR tree-optimization/79649
30561 * tree-loop-distribution.c (classify_partition): Give up on
30562 non-generic address space loads/stores.
30564 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
30566 * doc/loop.texi (Loop manipulation): Remove nonexistent
30567 tree_ssa_loop_version from the documentation.
30568 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
30570 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30573 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
30574 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
30575 * config/rs6000/rs6000.c: Include except.h.
30576 (rs6000_expand_split_stack_prologue): Call
30577 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
30579 2017-02-21 Martin Jambor <mjambor@suse.cz>
30582 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
30583 have been analyzed.
30585 2017-02-21 Martin Jambor <mjambor@suse.cz>
30587 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
30588 for backward compatibility only.
30589 * doc/invoke.texi (Option Summary): Remove all references to
30590 -fipa-cp-alignment.
30592 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
30596 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30598 * lra-constraints.c (curr_insn_transform): Handle
30599 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30601 2017-02-21 Martin Liska <mliska@suse.cz>
30603 * config/i386/i386.opt: Replace -masm-dialect with -masm.
30605 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
30607 PR translation/79638
30608 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
30610 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
30613 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
30614 (arm_function_ok_for_sibcall): Return false for an indirect call by
30615 descriptor if all the argument registers are used.
30616 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
30617 alignment of the function.
30619 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30621 PR tree-optimization/61441
30622 * simplify-rtx.c (simplify_const_unary_operation): For
30623 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
30624 the sNaN unmodified.
30626 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30628 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30629 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30630 instead of SYSTEM_HEADER_DIR.
30632 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
30633 Martin Liška <mliska@suse.cz>
30635 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
30636 Fix typos and grammar, use active voice, and clarify.
30638 2017-02-20 Marek Polacek <polacek@redhat.com>
30640 PR middle-end/79537
30641 * gimplify.c (gimplify_expr): Handle unused *&&L;.
30644 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
30646 2017-02-20 Jakub Jelinek <jakub@redhat.com>
30649 * config/i386/i386.c (ix86_expand_builtin): Handle
30650 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
30651 ix86_builtins_isa[fcode].isa as a requirement of those
30652 flags and any other flag in the bitmask.
30653 (ix86_init_mmx_sse_builtins): Use 0 instead of
30654 ~OPTION_MASK_ISA_64BIT as mask.
30655 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
30656 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
30657 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
30658 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
30660 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30663 * lra-constraints.c (split_reg): Check requested split mode
30664 is supported by the register.
30666 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30668 * lra-constraints.c (simplify_operand_subreg): Remove early
30671 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30674 * lra-constraints.c (curr_insn_transform): Tighten condition
30675 for converting SUBREG reloads from OP_OUT to OP_INOUT.
30677 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30680 * lra-constraints.c (curr_insn_transform): Handle
30681 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30683 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
30686 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
30688 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
30690 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
30693 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
30696 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30698 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
30701 2017-02-18 Jakub Jelinek <jakub@redhat.com>
30703 * final.c (last_columnnum, override_columnnum): New variables.
30704 (final_start_function): Set last_columnnum, pass it to begin_prologue
30705 hook and pass 0 to dwarf2out_begin_prologue.
30706 (final_scan_insn): Update override_columnnum. Pass last_columnnum
30707 to source_line debug hook.
30708 (notice_source_line): Compute last_columnnum and for debug_column_info
30709 return true on column changes.
30710 * debug.h (struct gcc_debug_hooks): Add column argument to
30711 source_line and begin_prologue hooks.
30712 (debug_nothing_int_charstar_int_bool): Remove prototype.
30713 (debug_nothing_int_int_charstar,
30714 debug_nothing_int_int_charstar_int_bool): New prototypes.
30715 (dwarf2out_begin_prologue): Add column argument.
30716 * debug.c (do_nothing_debug_hooks): Adjust source_line and
30717 begin_prologue hooks.
30718 (debug_nothing_int_charstar_int_bool): Remove.
30719 (debug_nothing_int_int_charstar,
30720 debug_nothing_int_int_charstar_int_bool): New functions.
30721 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
30722 through to dwarf2out_source_line.
30723 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
30724 (dwarf2out_source_line): Add column argument, emit it if requested.
30725 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
30727 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30728 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30729 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
30730 through to dwarf2out_begin_prologue.
30731 (vmsdbgout_source_line): Add column argument, pass it through to
30732 dwarf2out_source_line.
30733 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
30734 dbxout_source_line caller.
30735 (dbxout_source_line): Add column argument.
30737 * common.opt (gno-column-info, gcolumn-info): New options.
30738 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
30739 (check_die): Also test for multiple DW_AT_decl_column attributes.
30740 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30741 DW_AT_decl_column if requested.
30742 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30744 (gen_variable_die): Likewise.
30745 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30746 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30749 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30750 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30751 (ix86_handle_option): Handle OPT_m3dnowa.
30752 * doc/invoke.texi (-m3dnowa): Document.
30753 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30754 -m3dnowa instead of -m3dnow -march=athlon.
30757 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30758 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30760 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30763 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30764 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30765 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30766 generator for vsx_xxpermdi_<mode>_be.
30767 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
30768 force big-endian semantics.
30769 (vsx_xxpermdi_<mode>_be): New define_expand with same
30770 implementation as previous version of vsx_xxpermdi_<mode>.
30772 2017-02-17 Jakub Jelinek <jakub@redhat.com>
30774 PR tree-optimization/79327
30775 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
30776 variable, its initialization and use.
30778 2017-02-17 Julia Koval <julia.koval@intel.com>
30780 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
30781 (OPTION_MASK_ISA_PKU_UNSET): New.
30782 (ix86_handle_option): Handle -mrdpid.
30783 * config/i386/cpuid.h (bit_RDPID): New.
30784 * config/i386/driver-i386.c (host_detect_local_cpu):
30785 Detect RDPID feature.
30786 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
30787 * config/i386/i386-c.c (ix86_target_macros_internal):
30789 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
30790 (ix86_valid_target_attribute_inner_p): Add "rdpid".
30791 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
30792 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
30793 * config/i386/i386.md (define_insn "rdpid"): New.
30794 * config/i386/i386.opt Add -mrdpid.
30795 * config/i386/immintrin.h (_rdpid_u32): New.
30797 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
30799 PR rtl-optimization/79541
30800 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
30801 instead of transforming it into USE.
30803 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
30805 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
30806 If HONOR_SNANS (SFmode) force the input to a register.
30807 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
30808 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
30809 an frsp or similar insn.
30811 2017-02-17 Martin Liska <mliska@suse.cz>
30813 PR rtl-optimization/79577
30814 * params.def (selsched-max-sched-times): Increase minimum to 1.
30816 2017-02-17 Martin Liska <mliska@suse.cz>
30818 PR rtl-optimization/79574
30819 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30821 2017-02-17 Martin Liska <mliska@suse.cz>
30823 PR tree-optimization/79529
30824 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30825 ssa_defined_default_def_p to handle cases which are implicitly
30827 * tree-ssa.c (ssa_defined_default_def_p): New function.
30828 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30829 which are implicitly defined.
30830 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30832 2017-02-17 Richard Biener <rguenther@suse.de>
30834 PR middle-end/79576
30835 * params.def (max-ssa-name-query-depth): Limit to 10.
30837 2017-02-17 Richard Biener <rguenther@suse.de>
30839 PR tree-optimization/79552
30840 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30843 2017-02-17 Richard Biener <rguenther@suse.de>
30846 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30848 2017-02-17 Marek Polacek <polacek@redhat.com>
30850 PR middle-end/79536
30851 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30852 (fold_negate_expr): New wrapper.
30854 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30856 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30857 Correct terminology and de-emphasize pre-standard behavior.
30859 2017-02-16 Alan Modra <amodra@gmail.com>
30861 PR rtl-optimization/79286
30862 * ira.c (def_dominates_uses): New function.
30863 (update_equiv_regs): Don't create an equivalence for insns that
30864 may trap where the register def does not dominate the use.
30866 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30868 PR rtl-optimization/78127
30869 * lra.c (lra): Call lra_eliminate before finish the loop after
30872 2017-02-16 Richard Biener <rguenther@suse.de>
30874 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30876 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30877 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30878 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30879 (isl_val_int_from_wi): New function.
30880 (extract_affine_gmp): Rename to ...
30881 (extract_affine_wi): ... this, take a widest_int.
30882 (extract_affine_int): Just wrap extract_affine_wi.
30883 (add_param_constraints): Use isl_val_int_from_wi.
30884 (add_loop_constraints): Likewise, and extract_affine_wi.
30886 2017-02-15 Jeff Law <law@redhat.com>
30888 PR middle-end/79521
30889 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30890 ira_init_register_move_cost_if_necessary.
30892 2017-02-15 Martin Sebor <msebor@redhat.com>
30894 PR middle-end/32003
30895 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30896 removed in a prior commit.
30898 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30900 PR tree-optimization/79347
30901 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30902 counters during peeling.
30904 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30906 * Makefile.in (site.exp): Remove "set ISLVER".
30908 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30911 * real.c (real_from_integer): Call real_convert even for decimal.
30913 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30916 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30918 2017-02-14 Andrew Pinski <apinski@cavium.com>
30920 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30921 cores and change the partno/implementer to be correct.
30922 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30923 the 'B" as the implementer.
30924 * config/aarch64/aarch64-tune.md: Regenerate.
30926 2017-02-14 Carl Love <cel@us.ibm.com>
30928 * config/rs6000/rs6000.c: Add case statement entry to make the
30929 xvcvuxdsp built-in argument unsigned.
30930 * config/rs6000/vsx.md: Fix the source and return operand types so they
30931 match the instruction definitions from the ISA document. Fix typo
30932 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30935 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30938 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30939 member early_clobber_alts.
30940 * lra-lives.c (reg_early_clobber_p): New.
30941 (process_bb_lives): Use it.
30942 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30943 (debug_operand_data): Initialize early_clobber_alts.
30944 (setup_operand_alternative): Set up early_clobber_alts.
30945 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30946 alternatives to new_insn_reg.
30947 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30949 (lra_update_insn_regno_info): Pass the new arg.
30951 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30953 PR middle-end/79505
30954 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30955 (new_oacc_loop_raw): Don't clear already cleared fields.
30958 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30959 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30960 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30962 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30965 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30967 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30970 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30971 the extra instruction to the right place to store 128-bit constant
30974 2017-02-14 Martin Sebor <msebor@redhat.com>
30976 PR middle-end/79448
30977 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30978 warning for strings of unknown length.
30980 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30982 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30984 2017-02-14 Jeff Law <law@redhat.com>
30987 * ira-costs.c (scan_one_insn): Initialize register move costs
30988 for pseudos seen in USE/CLOBBER insns.
30990 PR tree-optimization/79095
30991 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30992 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30993 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30994 if the operands are known to be not equal, then the resulting range
30996 (intersect_ranges): If the new range is ~[0,0] and the old range is
30997 wide, then prefer ~[0,0].
30998 * tree-vrp.c (overflow_comparison_p_1): New function.
30999 (overflow_comparison_p): New function.
31000 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31001 if NAME is used in an overflow test.
31002 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31003 overflow check that can be expressed as an equality test, then adjust
31004 ops to be that equality test.
31006 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31008 * config/s390/s390-builtin-types.def: Remove flags argument.
31009 * config/s390/s390.c (s390_init_builtins): Likewise.
31011 2017-02-14 Martin Liska <mliska@suse.cz>
31013 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31014 vector. Fix trailing white spaces.
31016 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31018 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31021 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31023 PR rtl-optimization/68664
31024 * config/arm/arm.c (arm_sched_can_speculate_insn):
31025 New function. Declare prototype.
31026 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31028 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31030 PR rtl-optimization/68664
31031 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31033 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31035 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31037 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31038 max skip bytes for function, loop and jump.
31040 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31042 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31043 ABS_EXPR for gimple dump.
31045 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31048 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31050 PR tree-optimization/79408
31051 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31052 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31053 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31054 also if rhs1 is INTEGER_CST.
31056 2017-02-14 Richard Biener <rguenther@suse.de>
31058 PR middle-end/79432
31059 * tree-into-ssa.c (insert_phi_nodes): When the function can
31060 have abnormal edges rewrite SSA names with broken use-def
31061 dominance out of SSA and register them for PHI insertion.
31063 2017-02-13 Martin Sebor <msebor@redhat.com>
31065 PR middle-end/79496
31066 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31067 clearing info.nowrite flag when snprintf size argument is a range.
31069 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31071 * cprop.c (cprop_jump): Add missing space in string literal.
31072 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31073 (get_constraint_for_component_ref): Likewise.
31074 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31075 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31076 * lra-constraints.c (process_alt_operands): Likewise.
31077 * ipa-inline.c (inline_small_functions): Likewise.
31078 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31079 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31080 * trans-mem.c (diagnose_tm_1_op): Likewise.
31081 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31082 (grid_parallel_clauses_gridifiable): Likewise.
31084 * config/nvptx/mkoffload.c (process): Add space in between
31087 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31088 "MOD4_SSE_REGS" and "ALL_REGS".
31090 * spellcheck.c (test_data): Add , in between "foo" and "food".
31092 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31095 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31096 boundary crossing check and subsequent code generation agree.
31098 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31100 * config/aarch64/aarch64.c (has_memory_op): Delete.
31101 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31104 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31106 PR rtl-optimization/79388
31107 PR rtl-optimization/79450
31108 * combine.c (distribute_notes): When removing TEM_INSN for which
31109 corresponding dest has last value recorded, invalidate that last
31112 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31114 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31115 of explicit '@'. Add missing assembly comment marker on branch costs
31118 2017-02-13 Nathan Sidwell <nathan@acm.org>
31120 * gengtype-lex.l (<in_struct>): Add '/'.
31122 2017-02-13 Martin Liska <mliska@suse.cz>
31125 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31127 2017-02-13 Richard Biener <rguenther@suse.de>
31129 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31131 * configure: Re-generate.
31132 * config.in: Likewise.
31133 * graphite-dependences.c: Simplify as if
31134 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31135 * graphite-isl-ast-to-gimple.c: Likewise.
31136 * graphite-optimize-isl.c: Likewise.
31137 * graphite-poly.c: Likewise.
31138 * graphite-sese-to-poly.c: Likewise.
31139 * graphite.h: Likewise.
31140 * toplev.c: Include isl/version.h and use isl_version () for
31141 printing the ISL version.
31142 * doc/install.texi: Update ISL requirement.
31144 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31146 * doc/standards.texi (Standards): Update reference to
31149 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31151 * doc/extend.texi (Named Address Spaces): sourceware.org now
31153 * doc/install.texi (Binaries): Ditto.
31156 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31158 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31159 standard terminology "stringize"/"stringizing" throughout.
31160 * doc/cppinternals.texi: Likewise.
31162 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31164 * doc/extend.texi: Fix some spelling mistakes and typos.
31165 * doc/invoke.texi: Likewise.
31167 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31170 * params.def (inline-min-speedup) Change from 10 to 8.
31172 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31174 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31177 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31180 * ipa-inline-analysis.c (get_minimal_bb): New function.
31181 (record_modified): Use it.
31182 (remap_edge_change_prob): Handle also ancestor functions.
31184 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31186 * doc/contrib.texi (Contributors): Remove broken link into
31187 the Mauve CVS repository.
31189 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31191 PR middle-end/79454
31192 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31193 result computation whenever lhs doesn't have vector mode, not
31194 just when it has BLKmode.
31196 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31198 * doc/makefile.texi (profiledbootstrap): Refer to the
31199 installation instructions only in textual form.
31201 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31204 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31206 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31208 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31209 (Specific): Update mingw-w64 reference.
31211 (Specific): Remove broken link to Renesas RX processor.
31213 2017-02-10 Richard Biener <rguenther@suse.de>
31215 * toplev.c (process_options): Do not mention obsolete graphite
31216 options when printing sorry message about missing graphite support.
31217 Mention -floop-nest-optimize.
31219 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31221 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31222 (vtst_p16): Likewise.
31223 (vtstq_p8): Likewise.
31224 (vtstq_p16): Likewise.
31226 (vtstq_p64): Likewise.
31227 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31228 (vset_lane_p64): New.
31229 (vsetq_lane_p64): New.
31231 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31233 PR tree-optimization/79411
31234 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31235 stmt operands are SSA_NAMEs used in abnormal phis.
31236 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31239 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31242 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31245 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31247 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31249 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31251 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31255 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31256 not arbitrary TREE_CONSTANT.
31259 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31260 "omp declare target link" attribute unless is_global_var.
31261 * omp-offload.c (find_link_var_op): Likewise.
31263 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31264 Chung-Lin Tang <cltang@codesourcery.com>
31266 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31268 (gimplify_adjust_omp_clauses): Don't delete TILE.
31269 (gimplify_omp_for): Deal with TILE.
31270 * internal-fn.c (expand_GOACC_TILE): New function.
31271 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31273 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31274 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31276 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31277 avoid DIV for outermost collapse var.
31278 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31279 Remove out of date comments, fix whitespace.
31280 * omp-general.c (omp_extract_for_data): Deal with tiling.
31281 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31282 adjust OLF_DIM_BASE value.
31283 (struct omp_for_data): Add tiling field.
31284 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31285 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31286 for auto loops. Remove default auto determining, moved to
31287 oacc_loop_fixed_partitions.
31288 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31289 stmts, add e_mask field.
31290 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31291 (oacc_thread_numbers): Use oacc_dim_call.
31292 (oacc_xform_tile): New.
31293 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31294 (finish_oacc_loop): Adjust for ifns vector.
31295 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31296 add case for GOACC_TILE fns.
31297 (oacc_loop_xform_loop): Delete.
31298 (oacc_loop_process): Iterate over call list directly, and add
31299 handling for GOACC_TILE fns.
31300 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31302 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31303 vector partitioning to outer loops. Assign 2 partitions to loops
31304 when available. Add TILE handling.
31305 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31306 (execite_oacc_device_lower): Process GOACC_TILE fns,
31307 ignore unknown specs.
31308 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31309 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31310 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31312 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31314 * configure.ac (ACX_BUGURL): Update.
31315 * configure: Regenerate.
31317 2017-02-09 Richard Biener <rguenther@suse.de>
31319 PR tree-optimization/69823
31320 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31321 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31323 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31325 * config/arc/arc-c.def: Add __NPS400__ definition.
31326 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31327 (TARGET_NPS400): Define.
31329 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31331 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31333 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31334 pointer, arch_info.
31335 (arc_cpu_types): Fill the arch_info field with a pointer into the
31336 arc_arch_types table.
31337 (arc_selected_cpu): Declare.
31338 * config/arc/arc.c (arc_selected_cpu): Make global.
31339 (arc_selected_arch): Delete.
31340 (arc_base_cpu): Delete.
31341 (arc_override_options): Remove references to deleted variables,
31342 update access to arch information.
31343 (ARC_OPT): Update access to arch information.
31344 (ARC_OPTX): Likewise.
31345 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31346 (TARGET_ARC600): Update access to arch information.
31347 (TARGET_ARC601): Likewise.
31348 (TARGET_ARC700): Likewise.
31349 (TARGET_EM): Likewise.
31350 (TARGET_HS): Likewise.
31351 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31354 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31357 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31358 condition/operands for integer GE/LE/GEU/LEU operations.
31360 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31362 PR translation/79397
31363 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31366 2017-02-08 Martin Jambor <mjambor@suse.cz>
31369 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31370 whether allocation happened.
31371 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31372 nothing was allocated.
31374 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31376 PR tree-optimization/79408
31377 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31378 constant, but SSA_NAME with a known integer range, use the minimum
31379 of that range instead of op1 to determine if modulo can be replaced
31380 with its first operand.
31382 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31384 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31386 2017-02-08 Richard Biener <rguenther@suse.de>
31388 PR tree-optimization/71824
31389 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31390 Check all loops contained in the merged region.
31392 2017-02-07 Andrew Pinski <apinski@cavium.com>
31394 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31396 2017-02-07 Andrew Pinski <apinski@cavium.com>
31398 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31399 (thunderxt88): Likewise.
31400 (thunderxt81): Disable LSE and change v8.1 to v8.
31401 (thunderxt83): Likewise.
31403 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31404 Richard Biener <rguenther@suse.de>
31406 PR middle-end/79399
31407 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31408 type from int to size_t.
31409 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31411 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31413 PR rtl-optimization/79386
31414 * cprop.c (bypass_conditional_jumps): Initialize
31415 bypass_last_basic_block already before splitting bbs after
31416 unconditional traps...
31417 (bypass_conditional_jumps): ... rather than here.
31420 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31421 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31422 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31423 fix -masm=intel patterns.
31425 2017-02-07 Richard Biener <rguenther@suse.de>
31427 PR tree-optimization/79256
31428 PR middle-end/79278
31429 * builtins.c (get_object_alignment_2): Use min_align_of_type
31430 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31431 and ADJUST_FIELD_ALIGN.
31433 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31435 * doc/tm.texi: Regenerate.
31436 * stor-layout.c (layout_decl): Adjust.
31437 (update_alignment_for_field): Likewise.
31438 (place_field): Likewise.
31439 (min_align_of_type): Likewise.
31440 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31441 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31442 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31443 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31444 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31445 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31446 * config/i386/i386.c (x86_field_alignment): Likewise.
31447 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31448 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31449 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31450 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31451 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31452 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31456 2017-01-30 Richard Biener <rguenther@suse.de>
31458 PR tree-optimization/79256
31459 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31460 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31463 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31465 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31466 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31467 builtins to SImode and emit a zero-extend, if necessary.
31469 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31471 * docs/invoke.texi (RISC-V Options): Alphabetize.
31473 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31475 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31478 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31480 * config/riscv/riscv.c: New file.
31481 * common/config/riscv/riscv-common.c: Likewise.
31482 * config.gcc: Likewise.
31483 * config/riscv/constraints.md: Likewise.
31484 * config/riscv/elf.h: Likewise.
31485 * config/riscv/generic.md: Likewise.
31486 * config/riscv/linux.h: Likewise.
31487 * config/riscv/multilib-generator: Likewise.
31488 * config/riscv/peephole.md: Likewise.
31489 * config/riscv/pic.md: Likewise.
31490 * config/riscv/predicates.md: Likewise.
31491 * config/riscv/riscv-builtins.c: Likewise.
31492 * config/riscv/riscv-c.c: Likewise.
31493 * config/riscv/riscv-ftypes.def: Likewise.
31494 * config/riscv/riscv-modes.def: Likewise.
31495 * config/riscv/riscv-opts.h: Likewise.
31496 * config/riscv/riscv-protos.h: Likewise.
31497 * config/riscv/riscv.h: Likewise.
31498 * config/riscv/riscv.md: Likewise.
31499 * config/riscv/riscv.opt: Likewise.
31500 * config/riscv/sync.md: Likewise.
31501 * config/riscv/t-elf-multilib: Likewise.
31502 * config/riscv/t-linux: Likewise.
31503 * config/riscv/t-linux-multilib: Likewise.
31504 * config/riscv/t-riscv: Likewise.
31505 * configure.ac: Likewise.
31506 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31507 Waterman as RISC-V maintainers.
31508 * doc/install.texi: Add RISC-V entries.
31509 * doc/invoke.texi: Add RISC-V options section.
31510 * doc/md.texi: Add RISC-V constraints section.
31511 * configure: Regenerated.
31513 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31516 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31517 false values to be constant vectors with all 0 or all 1 bits set.
31518 (vcondu<mode><mode>): Likewise.
31519 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31521 (fpmask_comparison_operator): Update comment.
31522 (vecint_comparison_operator): New predicate.
31523 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31524 vector conditionals when the true and false values are constant
31525 vectors with all 0 bits or all 1 bits set.
31527 2017-02-06 Martin Sebor <msebor@redhat.com>
31529 PR tree-optimization/79376
31530 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31532 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31534 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31535 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31536 to simplify split condition.
31538 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31540 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31541 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31544 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31546 PR rtl-optimization/68664
31547 * target.def (can_speculate_insn): New hook.
31548 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31549 * doc/tm.texi: Regenerate.
31550 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31551 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31552 (rs6000_sched_can_speculate_insn): New function.
31554 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31556 PR tree-optimization/79284
31557 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
31558 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
31559 vectorizable_mask_load_store, vectorizable_operation,
31560 vect_is_simple_cond, get_same_sized_vectype): Use it instead
31561 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
31562 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
31563 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
31564 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
31565 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31566 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
31567 is_gimple_assign (stmt). Replace another such test with
31568 is_gimple_assign (stmt).
31570 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31573 * config/avr/avr.c (rtl-iter.h): Include it.
31574 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
31575 (avr_legitimate_combined_insn): ...and implementation.
31577 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
31579 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
31580 * config/s390/s390.c (s390_const_operand_ok)
31581 (s390_canonicalize_comparison, s390_extract_part)
31582 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
31583 (s390_contiguous_bitmask_p, s390_rtx_costs)
31584 (legitimize_pic_address): Likewise.
31585 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
31586 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
31587 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
31588 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
31589 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
31591 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31593 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
31594 REGNO($0) == REGNO($1).
31596 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31598 * config/s390/linux.h(SIZE_TYPE): Add comment.
31600 2017-02-06 Julian Brown <julian@codesourcery.com>
31601 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31602 Virendra Pathak <virendra.pathak@broadcom.com>
31604 * config/aarch64/aarch64-cores.def: Change the scheduler
31606 * config/aarch64/aarch64.md: Include thunderx2t99.md.
31607 * config/aarch64/thunderx2t99.md: New file.
31609 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31611 * doc/standards.texi (Go Language): Update link to language
31614 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31616 * tree-eh.c (lower_resx): Sanitize profile.
31617 (cleanup_empty_eh_move_lp): Likewise.
31619 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31622 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
31624 * cfgloopmanip.h (loop_version): Update prototype.
31625 * modulo-sched.c (sms_schedule): Update call of loop_version.
31626 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
31627 * tree-parloops.c (gen_parallel_loop): Likewise.
31628 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
31629 * tree-ssa-loop-split.c (split_loop): Likewise.
31630 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
31631 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
31633 2017-02-05 Martin Liska <mliska@suse.cz>
31636 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
31638 (print_operand_address): Initialize a struct to zero.
31640 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31642 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
31643 garbage collector only in textual form.
31645 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31647 * doc/extend.texi (x86 specific memory model extensions for
31648 transactional memory): Simplify a phrase.
31650 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
31653 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
31654 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
31655 (atomic_storedi_1): Likewise.
31657 2017-02-04 Jakub Jelinek <jakub@redhat.com>
31659 PR tree-optimization/79338
31660 * tree-parloops.c (gather_scalar_reductions): Don't call
31661 vect_analyze_loop_form for loop->inner before destroying loop's
31664 2017-02-03 Martin Sebor <msebor@redhat.com>
31666 PR tree-optimization/79327
31667 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
31668 when precision has resulted in leading zeros.
31669 (format_integer): Adjust the likely counter to assume an unknown
31670 argument that may be zero is non-zero.
31672 2017-02-03 Jason Merrill <jason@redhat.com>
31675 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
31676 avoid copying non-taken branch.
31678 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31680 PR tree-optimization/79340
31681 * tree-vect-loop.c (vectorizable_reduction): Release
31682 vec_defs elements after safe_splicing them into other vectors.
31685 PR tree-optimization/79327
31686 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
31687 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
31689 (format_integer): Use wide_int_to_tree instead of build_int_cst
31690 + to_?hwi. If argmin is NULL, just set argmin and argmax to
31691 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
31692 of shortest and longest sequence.
31694 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
31696 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
31697 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
31699 2017-02-03 Walter Lee <walt@tilera.com>
31702 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
31703 after initial stackframe link reg save.
31704 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
31706 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31709 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
31710 wu for stxssp alternative.
31712 2017-02-03 Martin Sebor <msebor@redhat.com>
31714 PR tree-optimization/79352
31715 * gimple-fold.c (get_range_strlen): Add argument.
31716 (get_range_strlen): Change return type to bool.
31717 (get_maxval_strlen): Pass in a dummy argument.
31718 * gimple-fold.h (get_range_strlen): Change return type to bool.
31719 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
31720 * tree.h (array_at_struct_end_p): Add argument.
31721 * tree.c (array_at_struct_end_p): Handle it.
31723 2017-02-03 Martin Liska <mliska@suse.cz>
31726 * multiple_target.c (create_dispatcher_calls): Redirect edge
31727 from a caller of a dispatcher.
31728 (expand_target_clones): Make the clones local.
31729 (ipa_target_clone): Do both target clones and resolvers.
31730 (ipa_dispatcher_calls): Remove the pass.
31731 (pass_dispatcher_calls::gate): Likewise.
31732 (make_pass_dispatcher_calls): Likewise.
31733 * passes.def (pass_target_clone): Put as very first IPA early
31736 2017-02-03 Martin Liska <mliska@suse.cz>
31738 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
31739 in case of a function with ifunc attribute.
31741 2017-02-03 Martin Liska <mliska@suse.cz>
31743 * cgraph.c (cgraph_node::dump): Dump function version info.
31744 * symtab.c (symtab_node::dump_base): Add missing new line.
31746 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31748 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31749 (ifcombine_ifandif): Use it.
31751 2017-02-03 Martin Liska <mliska@suse.cz>
31753 * doc/invoke.texi: Document default value for
31754 use-after-scope-direct-emission-threshold.
31756 2017-02-03 Martin Liska <mliska@suse.cz>
31758 PR tree-optimization/79339
31759 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31760 (format_floating): Likewise.
31762 2017-02-03 Martin Liska <mliska@suse.cz>
31765 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31767 (ipa_node_params_t::remove): Likewise.
31768 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
31769 initialization from removed ipa_node_params_t::insert.
31770 (ipa_node_params::~ipa_node_params): Move from removed
31771 ipa_node_params_t::release.
31772 * symbol-summary.h (symbol_summary::m_released): New member.
31773 Do not release a summary twice. Do not allow to call finalizer
31774 for types of a summary that live in GGC memory.
31776 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31778 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
31781 2017-02-02 Martin Sebor <msebor@redhat.com>
31783 PR middle-end/79275
31784 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
31785 (format_string): Tighten up the range of output for non-constant
31786 strings and correct the expected range for wide non-constant strings.
31788 2017-02-02 Martin Sebor <msebor@redhat.com>
31790 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
31792 PR middle-end/32003
31793 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
31795 (-fdump-tree-@var): Add to index and document how to come up
31796 with pass-specific option and dump file names.
31797 (-fdump-passes): Clarify where to look for output.
31799 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31801 PR middle-end/77445
31802 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
31803 statistics of the analyzed path; allow threading for speed when
31804 any of BBs along the path are optimized for speed.
31806 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
31808 PR middle-end/78468
31809 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
31810 settings of the virtual registers.
31813 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
31815 * explow.c (get_dynamic_stack_size): Take known alignment of stack
31816 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31819 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31821 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31822 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31824 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31826 * config/s390/s390.md: Add missing comments with the expanded
31828 * config/s390/vector.md: Likewise.
31829 * config/s390/vx-builtins.md: Likewise.
31831 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31834 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31835 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31836 conditions on a single line.
31838 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31840 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31841 __S390_VX__ to __VX__.
31843 2017-02-01 Andrew Pinski <apinski@cavium.com>
31845 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31846 stmt_info to record_stmt_cost.
31847 (vect_get_known_peeling_cost): Pass stmt_info if known to
31849 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31850 cpu_vector_cost field into
31851 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31852 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31853 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31854 splitting of scalar_stmt_cost and vec_stmt_cost.
31855 (thunderx_vector_cost): Likewise.
31856 (cortexa57_vector_cost): LIkewise.
31857 (exynosm1_vector_cost): Likewise.
31858 (xgene1_vector_cost): Likewise.
31859 (thunderx2t99_vector_cost): Improve after the splitting of the two
31861 (aarch64_builtin_vectorization_cost): Update for the splitting of
31862 scalar_stmt_cost and vec_stmt_cost.
31864 2017-02-01 Torvald Riegel <triegel@redhat.com>
31865 Richard Henderson <rth@redhat.com>
31867 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31868 conditional on existance of a fast atomic load.
31869 * optabs-query.c (can_atomic_load_p): New function.
31870 * optabs-query.h (can_atomic_load_p): Declare it.
31871 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31872 no fast atomic load is available for the particular size of access.
31873 (expand_atomic_compare_and_swap): Likewise.
31874 (expand_atomic_load): Likewise.
31875 (expand_atomic_store): Likewise.
31876 (expand_atomic_fetch_op): Likewise.
31877 * testsuite/lib/target-supports.exp
31878 (check_effective_target_sync_int_128): Remove x86 because it provides
31879 no fast atomic load.
31880 (check_effective_target_sync_int_128_runtime): Likewise.
31882 2017-02-01 Richard Biener <rguenther@suse.de>
31884 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31885 (graphite_transform_loops): Provide opt-info for optimized nests.
31886 * tree-parloop.c (parallelize_loops): Provide opt-info for
31887 parallelized loops.
31889 2017-02-01 Richard Biener <rguenther@suse.de>
31891 PR middle-end/79315
31892 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31893 was not set before.
31895 2017-02-01 Richard Biener <rguenther@suse.de>
31897 PR tree-optimization/71824
31898 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31899 Verify the loops are valid in the merged SESE region.
31900 (scop_detection::can_represent_loop_1): Check analyzing the
31901 evolution of the number of iterations in the region succeeds.
31903 2017-01-31 Ian Lance Taylor <iant@golang.org>
31905 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31906 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31908 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31910 PR preprocessor/79210
31911 * input.c (get_substring_ranges_for_loc): Replace line_width
31912 assertion with error-handling.
31914 2017-01-31 Richard Biener <rguenther@suse.de>
31916 PR tree-optimization/77318
31917 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31918 (create_pw_aff_from_tree): Take loop parameter.
31919 (add_condition_to_pbb): Pass loop of the condition to
31920 create_pw_aff_from_tree.
31922 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31924 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31925 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31927 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31931 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31933 (convert_int_to_float128): Likewise.
31934 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31935 (convert_int_to_float128): Likewise.
31936 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31937 (UNSPEC_IEEE128_CONVERT): Likewise.
31938 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31939 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31940 Use local variables for IBM extended format.
31941 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31942 (fix_trunc<mode>si2_fprs): Likewise.
31943 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31944 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31945 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31946 to know that we can now have integers of all sizes in vector
31948 (fix<uns>_<mode>di2_hw): Likewise.
31949 (float<uns>_<mode>si2_hw): Likewise.
31950 (fix_<mode>si2_hw): Likewise.
31951 (fixuns_<mode>si2_hw): Likewise.
31952 (float<uns>_<mode>di2_hw): Likewise.
31953 (float_<mode>di2_hw): Likewise.
31954 (float_<mode>si2_hw): Likewise.
31955 (floatuns_<mode>di2_hw): Likewise.
31956 (floatuns_<mode>si2_hw): Likewise.
31957 (xscvqp<su>wz_<mode>): Delete, no longer used.
31958 (xscvqp<su>dz_<mode>): Likewise.
31959 (xscv<su>dqp_<mode>): Likewise.
31960 (ieee128_mfvsrd_64bit): Likewise.
31961 (ieee128_mfvsrd_32bit): Likewise.
31962 (ieee128_mfvsrwz): Likewise.
31963 (ieee128_mtvsrw): Likewise.
31964 (ieee128_mtvsrd_64bit): Likewise.
31965 (ieee128_mtvsrd_32bit): Likewise.
31967 2017-01-31 Martin Liska <mliska@suse.cz>
31970 * ipa-prop.c (ipa_free_all_node_params): Call release method
31971 instead of ~sumbol_summary to not to trigger double times
31974 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31976 PR tree-optimization/71691
31977 * bitmap.h (class auto_bitmap): New.
31978 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31979 is_maybe_undefined instead of ssa_undefined_value_p.
31981 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31983 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31984 __S390_ARCH_LEVEL__ to __ARCH__.
31986 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31988 PR tree-optimization/79267
31989 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31990 if should_remove_lhs_p is true.
31992 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31995 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31996 (add_alignment_attribute): New.
31997 (base_type_die): Add alignment attribute.
31998 (subrange_type_die): Likewise.
31999 (modified_type_die): Likewise.
32000 (gen_array_type_die): Likewise.
32001 (gen_descr_array_type_die: Likewise.
32002 (gen_enumeration_type_die): Likewise.
32003 (gen_subprogram_die): Likewise.
32004 (gen_variable_die): Likewise.
32005 (gen_field_die): Likewise.
32006 (gen_ptr_to_mbr_type_die): Likewise.
32007 (gen_struct_or_union_type_die): Likewise.
32008 (gen_subroutine_type_die): Likewise.
32009 (gen_typedef_die): Likewise.
32010 (base_type_cmp): Compare alignment attribute.
32012 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32015 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32016 (setb_unsigned) New pattern for setb with CCUNS.
32017 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32018 subfc./subfe sequence to avoid overflow problems. Generate a
32019 shorter sequence with cmpld/setb for power9.
32020 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32021 for generating subfc. instruction.
32022 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32023 now uses this instruction.
32025 2017-01-30 Ian Lance Taylor <iant@google.com>
32028 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32029 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32031 2017-01-30 Martin Sebor <msebor@redhat.com>
32033 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32034 Move constant to the right of a relational operator.
32035 (get_mpfr_format_length, format_character, format_string): Ditto.
32036 (should_warn_p, maybe_warn): Same.
32038 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32040 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32043 * asan.c (get_translation_unit_decl): Remove function.
32044 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32046 2017-01-30 Martin Liska <mliska@suse.cz>
32048 PR gcov-profile/79259
32049 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32050 -fprofile-generate.
32052 2017-01-30 Martin Liska <mliska@suse.cz>
32055 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32056 Initialize variables with NULL value.
32058 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32061 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32063 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32065 2017-01-30 Richard Biener <rguenther@suse.de>
32067 PR tree-optimization/79276
32068 * tree-vrp.c (process_assert_insertions): Properly adjust common
32069 when removing a duplicate.
32071 2017-01-30 Richard Biener <rguenther@suse.de>
32073 PR tree-optimization/79256
32074 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32075 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32077 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32079 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32082 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32083 ("*r<noxa>sbg_<mode>_sll_bitmask")
32084 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32085 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32086 Use contiguous_bitmask_nowrap_operand.
32088 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32091 * config/rs6000/altivec.h (vec_xl): Revise #define.
32092 (vec_xst): Likewise.
32094 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32096 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32098 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32100 PR rtl-optimization/79194
32101 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32102 traps before call to bypass_conditional_jumps.
32104 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32106 PR tree-optimization/71374
32107 * lra-constraints.c (check_conflict_input_operands): New.
32108 (match_reload): Use it.
32110 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32113 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32114 account to calculate conflict_set.
32116 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32118 PR rtl-optimization/78559
32119 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32120 other_insn in combine.
32122 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32124 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32125 uint16_type_node for BT_UINT16.
32127 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32129 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32130 "RTL Tests" to menu.
32131 (GIMPLE Tests): New node.
32132 (RTL Tests): New node.
32134 2017-01-27 Richard Biener <rguenther@suse.de>
32136 PR tree-optimization/79245
32137 * tree-loop-distribution.c (distribute_loop): Apply cost
32138 modeling also to detected patterns.
32140 2017-01-27 Richard Biener <rguenther@suse.de>
32142 PR tree-optimization/71433
32143 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32144 (compare_assert_loc): New function.
32145 (process_assert_insertions): Sort and optimize assert locations
32146 to remove duplicates and push down identical assertions on
32147 edges to their destination block.
32149 2017-01-27 Richard Biener <rguenther@suse.de>
32151 PR tree-optimization/79244
32152 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32153 out SSA names even if abnormal.
32155 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32157 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32158 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32159 instead of MPFR_RNDN.
32161 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32164 * arm.c (arm_option_override): Don't call build_target_option_node
32165 until after doing all option overrides.
32166 (arm_valid_target_attribute_tree): Likewise.
32168 2017-01-27 Martin Liska <mliska@suse.cz>
32170 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32171 for {cd}tors and C++ {cd}tors.
32173 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32175 * config/s390/s390.md ("*setmem_long_and")
32176 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32178 2017-01-26 Martin Sebor <msebor@redhat.com>
32180 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32183 2017-01-26 Martin Sebor <msebor@redhat.com>
32185 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32186 HAVE_DFmode before using XFmode or DFmode.
32187 (parse_directive): Avoid using the z length modifier to avoid
32188 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32190 PR middle-end/78703
32191 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32192 to accept adjustment as an array.
32193 (get_int_range): New function.
32194 (struct directive): Make width and prec arrays.
32195 (directive::set_width, directive::set_precision): Call get_int_range.
32196 (format_integer, format_floating): Handle width and precision ranges.
32197 (format_string, parse_directive): Same.
32199 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32202 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32203 comdat_type_p set, just clone them, but keep the children in the
32207 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32208 which have direct callers with -fvar-tracking-assignments enabled
32210 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32211 inside of type units.
32213 2017-01-26 Martin Sebor <msebor@redhat.com>
32215 PR middle-end/78703
32216 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32218 (struct format_result): Replace number_chars, number_chars_min,
32219 and number_chars_max with a single member of struct result_range.
32221 (format_result::operator+=): Adjust.
32222 (struct fmtresult): Remove bounded. Handle likely and unlikely
32224 (fmtresult::adjust_for_width_or_precision): New function.
32225 (fmtresult:type_max_digits): New function.
32226 (bytes_remaining): Handle likely and unlikely counters.
32227 (min_bytes_remaining): Remove.
32228 (format_percent): Simplify.
32229 (format_integer, format_floating): Set likely and unlikely counters.
32230 (get_string_length, format_character, format_string): Same.
32231 (format_plain, should_warn_p): New function.
32232 (maybe_warn): Call should_warn_p. Update diagnostic messages
32233 and handle those for all directives, including plain strings.
32234 (format_directive): Handle likely and unlikely counters.
32235 Remove unnecessary quoting from diagnostics. Add an informational
32237 (add_bytes): Remove.
32238 (pass_sprintf_length::compute_format_length): Simplify.
32239 (try_substitute_return_value): Handle likely and unlikely counters.
32241 2017-01-26 Carl Love <cel@us.ibm.com>
32243 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32244 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32246 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32249 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32250 endianess for subregs into account.
32251 * lra-constraints.c (lra_constraints): Do risky transformations
32252 always on the first iteration.
32253 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32254 last_call_used_reg_set.
32255 (process_bb_lives): Define and use last_call_used_reg_set.
32256 * lra.c (lra): Always continue after lra_constraints on the first
32259 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32261 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32263 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32265 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32267 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32268 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32269 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32270 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32271 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32272 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32273 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32274 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32275 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32277 2017-01-26 Marek Polacek <polacek@redhat.com>
32280 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32281 for the third operand.
32283 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32285 PR middle-end/79236
32286 * omp-low.c (struct omp_context): Add simt_stmt field.
32287 (scan_omp_for): Return omp_context *.
32288 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32289 context to the _simt_ SIMD stmt.
32290 (lower_omp_for): For combined SIMD with sibling _simt_
32291 SIMD, make sure to use the same decls in _looptemp_
32292 clauses as in the sibling.
32294 2017-01-26 David Sherwood <david.sherwood@arm.com>
32296 PR middle-end/79212
32297 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32300 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32303 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32304 emit fld b; fld a; if possible.
32306 * brig-builtins.def: Update copyright years.
32307 * config/arm/arm_acle_builtins.def: Update copyright years.
32309 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32312 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32313 constraint instead of o for the stxsd instruction.
32315 2017-01-25 Carl Love <cel@us.ibm.com>
32317 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32318 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32320 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32322 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32324 2017-01-25 Richard Biener <rguenther@suse.de>
32326 PR tree-optimization/69264
32327 * target.def (vector_alignment_reachable): Improve documentation.
32328 * doc/tm.texi: Regenerate.
32329 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32331 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32332 earlier changes with respect to TYPE_USER_ALIGN.
32333 (vector_alignment_reachable_p): Likewise. Improve dumping.
32335 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32338 * config/arm/arm.md (xordi3): Force constant operand into a register
32341 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32343 * doc/invoke.texi (-fstore-merging): Correct default optimization
32344 levels at which it is enabled.
32345 (-O): Move -fstore-merging from list to...
32348 2017-01-25 Richard Biener <rguenther@suse.de>
32351 * omp-expand.c: Include debug.h.
32352 (expand_omp_taskreg): Make sure to generate early debug before
32353 outlining anything from a function.
32354 (expand_omp_target): Likewise.
32355 (grid_expand_target_grid_body): Likewise.
32357 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32360 * asan.c (get_translation_unit_decl): New function.
32361 (asan_add_global): Extract modules file name from globals
32362 TRANSLATION_UNIT_DECL name.
32364 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32367 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32368 for long calls with APCS frame and VFP.
32370 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32372 * cfg.c (original_copy_tables_initialized_p): New function.
32373 * cfg.h (original_copy_tables_initialized_p): New decl.
32374 * cfgrtl.c (relink_block_chain): Guard the call to
32375 free_original_copy_tables with a call to
32376 original_copy_tables_initialized_p.
32377 * cgraph.h (symtab_node::native_rtl_p): New decl.
32378 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32379 (symtab_node::needed_p): Don't assert for early assembly output
32380 for __RTL functions.
32381 (cgraph_node::finalize_function): Set "force_output" for __RTL
32383 (cgraph_node::analyze): Bail out early for __RTL functions.
32384 (analyze_functions): Update assertion to support __RTL functions.
32385 (cgraph_node::expand): Bail out early for __RTL functions.
32386 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32388 * function.h (struct function): Update comment for field
32390 * gimple-expr.c: Include "tree-pass.h".
32391 (gimple_has_body_p): Return false for __RTL functions.
32392 * Makefile.in (OBJS): Add run-rtl-passes.o.
32393 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32395 (gcc::pass_manager::get_clean_slate): New accessor.
32396 * passes.c: Include "insn-addr.h".
32397 (should_skip_pass_p): Add logging. Update logic for running
32398 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32399 property-provider override so it is only done for gimple passes.
32401 (skip_pass): New function.
32402 (execute_one_pass): Call skip_pass when skipping passes.
32403 * read-md.c (md_reader::read_char): Support filtering
32404 the input to a subset of line numbers.
32405 (md_reader::md_reader): Initialize fields
32406 m_first_line and m_last_line.
32407 (md_reader::read_file_fragment): New function.
32408 * read-md.h (md_reader::read_file_fragment): New decl.
32409 (md_reader::m_first_line): New field.
32410 (md_reader::m_last_line): New field.
32411 * read-rtl-function.c (function_reader::create_function): Only
32412 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32413 curr_properties. Set DECL_INITIAL to a dummy block.
32414 (read_rtl_function_body_from_file_range): New function.
32415 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32417 * run-rtl-passes.c: New file.
32418 * run-rtl-passes.h: New file.
32420 2017-01-24 Jeff Law <law@redhat.com>
32422 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32425 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32427 PR tree-optimization/79159
32428 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32429 (record_nonwrapping_iv): Improve boundary using above function if no
32430 value range information.
32432 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32433 Martin Jambor <mjambor@suse.cz>
32435 * brig-builtins.def: New file.
32436 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32437 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32438 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32439 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32440 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32441 * builtin-types.def (BT_INT8): New.
32442 (BT_INT16): Likewise.
32443 (BT_UINT8): Likewise.
32444 (BT_UINT16): Likewise.
32445 (BT_FN_ULONG): Likewise.
32446 (BT_FN_UINT_INT): Likewise.
32447 (BT_FN_UINT_ULONG): Likewise.
32448 (BT_FN_UINT_LONG): Likewise.
32449 (BT_FN_UINT_PTR): Likewise.
32450 (BT_FN_ULONG_PTR): Likewise.
32451 (BT_FN_INT8_FLOAT): Likewise.
32452 (BT_FN_INT16_FLOAT): Likewise.
32453 (BT_FN_UINT32_FLOAT): Likewise.
32454 (BT_FN_UINT16_FLOAT): Likewise.
32455 (BT_FN_UINT8_FLOAT): Likewise.
32456 (BT_FN_UINT64_FLOAT): Likewise.
32457 (BT_FN_UINT16_UINT32): Likewise.
32458 (BT_FN_UINT32_UINT16): Likewise.
32459 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32460 (BT_FN_INT_PTR_INT): Likewise.
32461 (BT_FN_UINT_PTR_UINT): Likewise.
32462 (BT_FN_LONG_PTR_LONG): Likewise.
32463 (BT_FN_ULONG_PTR_ULONG): Likewise.
32464 (BT_FN_VOID_UINT64_UINT64): Likewise.
32465 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32466 (BT_FN_INT8_INT8_INT8): Likewise.
32467 (BT_FN_INT16_INT16_INT16): Likewise.
32468 (BT_FN_INT_INT_INT): Likewise.
32469 (BT_FN_UINT_FLOAT_UINT): Likewise.
32470 (BT_FN_FLOAT_UINT_UINT): Likewise.
32471 (BT_FN_ULONG_UINT_UINT): Likewise.
32472 (BT_FN_ULONG_UINT_PTR): Likewise.
32473 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32474 (BT_FN_UINT_UINT_UINT): Likewise.
32475 (BT_FN_VOID_UINT_PTR): Likewise.
32476 (BT_FN_UINT_UINT_PTR: Likewise.
32477 (BT_FN_UINT32_UINT64_PTR): Likewise.
32478 (BT_FN_INT_INT_UINT_UINT): Likewise.
32479 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32480 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32481 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32482 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32483 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32484 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32485 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32486 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32487 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32488 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32489 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32490 * doc/frontends.texi: List BRIG FE.
32491 * doc/install.texi (Testing): Add BRIG tesring requirements.
32492 * doc/invoke.texi (Overall Options): Mention BRIG.
32493 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32495 2017-01-24 Richard Biener <rguenther@suse.de>
32497 PR translation/79208
32498 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32500 2017-01-24 Martin Jambor <mjambor@suse.cz>
32503 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32504 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32505 and known_contexts.
32507 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32509 PR middle-end/79123
32510 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32511 casts from signed to unsigned really don't have a range.
32513 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32515 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32516 GMP_RNDx for compatiblity.
32518 2017-01-24 Martin Liska <mliska@suse.cz>
32521 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32522 that would prevent us to call alloca with -1 as argument.
32524 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32526 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32527 Avoid -Wformat-security warning.
32529 2017-01-23 Andrew Pinski <apinski@cavium.com>
32531 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32534 2017-01-23 Martin Sebor <msebor@redhat.com>
32536 PR middle-end/78703
32537 * gimple-ssa-sprintf.c (warn_level): New global.
32538 (format_integer): Use it here and throughout the rest of the file.
32539 Use the same switch to compute sign as base.
32540 (maybe_warn): New function.
32541 (format_directive): Factor out warnings into maybe_warn.
32542 Add debugging output. Use warn_level.
32543 (add_bytes): Use warn_level.
32544 (pass_sprintf_length::compute_format_length): Add debugging output.
32545 (try_substitute_return_value): Same.
32546 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32548 PR middle-end/78703
32549 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32550 (struct fmtresult, format_integer, format_floating): Adjust.
32551 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32552 (get_string_length, format_string,format_directive): Same.
32553 (pass_sprintf_length::compute_format_length): Same.
32554 (try_substitute_return_value): Simplify slightly.
32556 PR middle-end/78703
32557 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
32558 (fmtresult::operator+=): Outlined.
32559 (struct fmtresult): Add ctors.
32560 (struct conversion_spec): Rename...
32561 (struct directive): ...to this. Add and remove data members.
32562 (directive::set_width, directive::set_precision): New functions.
32563 (format_percent): Use fmtresult ctor.
32564 (get_width_and_precision): Remove.
32565 (format_integer): Make naming changes. Avoid computing width and
32567 (format_floating): Same. Adjust indentation.
32568 (format_character, format_none): New functions.
32569 (format_string): Moved character handling to format_character.
32570 (format_directive): Remove arguments, change return type.
32571 (parse_directive): New function.
32572 (pass_sprintf_length::compute_format_length): Move directive
32573 parsing to parse_directive.
32575 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32577 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
32578 (assign_assembler_name_if_needed): ... this.
32579 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
32580 (assign_assembler_name_if_needed): ... this.
32581 (free_lang_data_in_cgraph): Adjust callers.
32582 * cgraphunit.c (cgraph_node::analyze): Likewise.
32583 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
32586 2017-01-23 Richard Biener <rguenther@suse.de>
32588 PR tree-optimization/79088
32589 PR tree-optimization/79188
32590 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
32591 resetting loop bounds after last path deletion. Reset loop
32592 bounds of the target loop, make code match the comments.
32593 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
32594 Make sure loops need no fixups.
32596 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32598 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
32599 exponent support with double type for first argument.
32600 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
32601 type returned by __builtin_vec_extract_sig,
32602 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
32603 functions from "vector int" to "vector unsigned int" or from
32604 "vector long long int" to "vector unsigned long long int".
32605 Changed type returned by __builtin_vec_extract_exp,
32606 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
32607 functions from "vector int" to "vector unsigned int" or from
32608 "vector long long int" to "vector unsigned long long int".
32609 Changed return type of __builtin_vec_test_data_class,
32610 __builtin_vec_test_data_class_sp, and
32611 __builtin_vec_test_data_class_dp from "vector int" to
32612 "vector bool int" or from "vector long long int" to "vector bool
32613 long long int" and changed second argument type from "unsigned
32614 int" to "int". Added new overloaded function forms "vector float
32615 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
32616 "vector float __builtin_vec_insert_exp_sp (vector float, vector
32617 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
32618 double, vector unsigned long long int)" and "vector double
32619 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
32620 long int)". Changed return type of
32621 __builtin_scalar_test_data_class and
32622 __builtin_scalar_test_data_class_sp and
32623 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
32624 int" and changed second argument from "unsigned int" to "int".
32625 Changed type returned by __builtin_scalar_test_neg,
32626 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
32627 from "int" to "bool int". Added new overloaded function form
32628 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
32629 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
32630 exponent double-precision with floating point first argument.
32631 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
32632 documentation of scalar_test_data_class, scalar_test_neg,
32633 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
32634 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
32635 vec_test_data_class built-in functions to reflect refinements in
32636 their type signatures.
32638 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
32640 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
32642 (aarch64_elf_asm_destructor): Likewise.
32644 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
32646 PR rtl-optimization/78634
32647 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
32648 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
32649 * ifcvt.c (noce_try_cmove): Add missing cost check.
32651 PR rtl-optimization/71724
32652 * combine.c (if_then_else_cond): Look for situations where it is
32653 beneficial to undo the work of one of the recursive calls.
32655 2017-01-23 Bin Cheng <bin.cheng@arm.com>
32657 PR tree-optimization/70754
32658 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
32659 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
32660 combined stmt before it if not NULL.
32661 (combine_chains): Process refs reversely and compute dominance point
32664 2017-01-23 Martin Liska <mliska@suse.cz>
32666 PR tree-optimization/79196
32667 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
32668 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
32670 (strlen_optimize_stmt): Call the renamed function.
32672 2017-01-23 Michael Matz <matz@suse.de>
32674 PR tree-optimization/78384
32675 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
32677 2017-01-23 Richard Biener <rguenther@suse.de>
32679 PR tree-optimization/79186
32680 * tree-vrp.c (register_new_assert_for): Make sure we've seen
32681 both incoming edges before moving an assert.
32683 2017-01-23 Martin Jambor <mjambor@suse.cz>
32685 * ipa-prop.c (load_from_param_1): Removed.
32686 (load_from_unmodified_param): Bits from load_from_param_1 put back
32688 (load_from_param): Removed.
32689 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
32690 with stmt. Reverted back to use of load_from_unmodified_param.
32692 2017-01-23 Martin Jambor <mjambor@suse.cz>
32695 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
32696 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
32697 field a pointer to garbage collected vector, mark lattices and
32698 ipcp_orig_node with GTY((skip)).
32699 (ipa_get_param_count): Adjust to descriptors being a pointer.
32700 (ipa_get_param): Likewise.
32701 (ipa_get_type): Likewise.
32702 (ipa_get_param_move_cost): Likewise.
32703 (ipa_set_param_used): Likewise.
32704 (ipa_get_controlled_uses): Likewise.
32705 (ipa_set_controlled_uses): Likewise.
32706 (ipa_is_param_used): Likewise.
32707 (ipa_node_params_t): Move into garbage collector. New methods insert
32709 (ipa_node_params_sum): Annotate wth GTY(()).
32710 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
32712 (ipa_load_from_parm_agg): Adjust declaration.
32713 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
32714 * ipa-profile.c (ipa_profile): Likewise.
32715 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
32716 (ipa_populate_param_decls): Make descriptors parameter garbage
32718 (ipa_dump_param): Adjust to descriptors being a pointer.
32719 (ipa_alloc_node_params): Likewise.
32720 (ipa_initialize_node_params): Likewise.
32721 (load_from_param_1): Make descriptors parameter garbage collected.
32722 (load_from_unmodified_param): Likewise.
32723 (load_from_param): Likewise.
32724 (ipa_load_from_parm_agg): Likewise.
32725 (ipa_node_params::~ipa_node_params): Removed.
32726 (ipa_free_all_node_params): Remove call to delete operator.
32727 (ipa_node_params_t::insert): New.
32728 (ipa_node_params_t::remove): Likewise.
32729 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
32730 copy known_csts and known_contexts vectors.
32731 (ipa_read_node_info): Adjust to descriptors being a pointer.
32732 (ipcp_modif_dom_walker): Make m_descriptors field garbage
32734 (ipcp_transform_function): Make descriptors variable garbage
32737 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
32739 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32740 * config/i386/avx512dqintrin.h: Ditto.
32741 * config/i386/avx512fintrin.h: Ditto.
32742 * config/i386/i386.c: Handle new builtins.
32743 * config/i386/i386-builtin.def: Add new builtins.
32744 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32745 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32747 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32748 Martin Liska <mliska@suse.cz>
32750 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32751 * asan.c (asan_expand_poison_ifn): Support stores and use
32752 appropriate ASAN report function.
32753 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32754 * internal-fn.def (ASAN_POISON_USE): Declare.
32755 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32756 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32757 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32758 ASAN_POISON calls w/o LHS.
32759 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32760 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32761 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32762 * gimplify.c (asan_poison_variables): Add attribute
32763 use_after_scope_memory to variables that really needs to live
32765 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32766 having the attribute.
32768 2017-01-23 Martin Liska <mliska@suse.cz>
32770 * asan.c (create_asan_shadow_var): New function.
32771 (asan_expand_poison_ifn): Likewise.
32772 * asan.h (asan_expand_poison_ifn): New declaration.
32773 * internal-fn.c (expand_ASAN_POISON): Likewise.
32774 * internal-fn.def (ASAN_POISON): New builtin.
32775 * sanopt.c (pass_sanopt::execute): Expand
32776 asan_expand_poison_ifn.
32777 * tree-inline.c (copy_decl_for_dup_finish): Make function
32779 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
32780 * tree-ssa.c (is_asan_mark_p): New function.
32781 (execute_update_addresses_taken): Rewrite local variables
32782 (identified just by use-after-scope as addressable) into SSA.
32784 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32786 * doc/install.texi (Specific): opensource.apple.com uses https
32787 now. Remove trailing slash.
32789 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32791 * README.Portability: Remove note on an Irix compatibility issue.
32793 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
32795 * gcov.c (INCLUDE_ALGORITHM): Define.
32796 (INCLUDE_VECTOR): Define.
32797 No longer include <vector> and <algorithm> directly.
32799 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
32801 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
32803 * doc/invoke.texi (Code Gen Options): Ditto.
32805 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
32808 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
32810 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
32812 rtl-optimization/79125
32813 * cprop.c (local_cprop_pass): Handle cases where we make an
32814 unconditional trap.
32816 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32820 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32821 read from, for big endian.
32823 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32825 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32826 register pauth builtins for LP64 only.
32828 2017-01-20 Marek Polacek <polacek@redhat.com>
32831 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32834 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32836 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32838 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32839 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32840 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32842 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32845 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32846 in big-endian mode when they are not a single duplicated value.
32848 2017-01-20 Richard Biener <rguenther@suse.de>
32850 * BASE-VER: Bump to 7.0.1.
32852 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32854 * omp-low.c (omplow_simd_context): New struct. Use it...
32855 (lower_rec_simd_input_clauses): ...here and...
32856 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32857 references to idx, lane, max_vf, is_simt.
32859 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32861 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32864 2017-01-20 Martin Jambor <mjambor@suse.cz>
32866 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32867 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32869 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32870 (GTFILES): Rename hsa.c to hsa-common.c.
32871 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32872 * hsa-dump.c: Likewise.
32873 * hsa-gen.c: Likewise.
32874 * hsa-regalloc.c: Likewise.
32875 * ipa-hsa.c: Likewise.
32876 * omp-expand.c: Likewise.
32877 * omp-low.c: Likewise.
32878 * toplev.c: Likewise.
32880 2017-01-20 Marek Polacek <polacek@redhat.com>
32883 * doc/invoke.texi: Document -Wduplicated-branches.
32884 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32885 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32886 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32887 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32888 return 0 only when not OEP_LEXICOGRAPHIC.
32889 (fold_build_cleanup_point_expr): Use the expression
32890 location when building CLEANUP_POINT_EXPR.
32891 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32892 * tree.c (add_expr): Handle error_mark_node.
32894 2017-01-20 Martin Liska <mliska@suse.cz>
32897 * tree-profile.c (init_ic_make_global_vars): Do not call
32899 (gimple_init_gcov_profiler): Likewise.
32901 2017-01-20 Martin Liska <mliska@suse.cz>
32904 * cgraph.h (maybe_create_reference): Remove argument and
32906 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32908 * ipa-cp.c (create_specialized_node): Likewise.
32909 * symtab.c (symtab_node::maybe_create_reference): Handle
32910 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32912 2017-01-20 Martin Liska <mliska@suse.cz>
32914 * read-rtl-function.c (function_reader::create_function): Use
32915 build_decl instread of build_decl_stat.
32917 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32919 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32920 * config/i386/avx512dqintrin.h: Ditto.
32921 * config/i386/avx512fintrin.h: Ditto.
32922 * config/i386/i386-builtin-types.def: Add new types.
32923 * config/i386/i386.c: Handle new types.
32924 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32925 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32926 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32927 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32928 (__builtin_ia32_kshiftridi): New.
32929 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32931 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32935 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32936 define to rs6000_init_stack_protect_guard.
32937 (rs6000_init_stack_protect_guard): New function.
32939 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32940 Yunqiang Su <yunqiang.su@imgtec.com>
32942 * config.gcc (supported_defaults): Add madd4.
32943 (with_madd4): Add validation.
32944 (all_defaults): Add madd4.
32945 * config/mips/mips.opt (mmadd4): New option.
32946 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32948 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32950 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32951 (ISA_HAS_FUSED_MADD4): Likewise.
32952 * doc/invoke.texi (-mmadd4): Document the new option.
32953 * doc/install.texi (--with-madd4): Document the new option.
32955 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32957 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32958 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32959 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32960 (aarch64_init_pauth_hint_builtins): New.
32961 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32962 (aarch64_expand_builtin): Expand new builtins.
32964 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32966 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32967 * combine-stack-adj.c (no_unhandled_cfa): Handle
32968 REG_CFA_TOGGLE_RA_MANGLE.
32969 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32970 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32971 info for return address signing.
32972 (aarch64_expand_epilogue): Likewise.
32974 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32976 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32977 * config/aarch64/aarch64-protos.h
32978 (aarch64_return_address_signing_enabled): New declaration.
32979 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32981 (aarch64_expand_prologue): Sign return address before it's pushed onto
32983 (aarch64_expand_epilogue): Authenticate return address fetched from
32985 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32986 (aarch64_attributes): New function attributes for "sign-return-address".
32987 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32988 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32989 ("*do_return"): Generate combined instructions according to key index.
32990 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32991 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32993 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32994 * config/aarch64/aarch64.opt (msign-return-address=): New.
32995 * doc/extend.texi (AArch64 Function Attributes): Documents
32996 "sign-return-address=".
32997 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32999 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33001 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33002 overall option summary.
33004 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33006 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33007 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33008 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33009 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33011 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33013 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33014 -mpower9-minmax by default for -mcpu=power9.
33015 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33016 128-bit floating point.
33018 2017-01-20 Alan Modra <amodra@gmail.com>
33020 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33021 optimizing for size.
33023 2017-01-20 Alan Modra <amodra@gmail.com>
33026 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33027 for strcmp and strncmp from corresponding builtin decl.
33029 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33031 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33032 instead of i386/rtems-64.h.
33033 * config/i386/rtems-64.h: Remove.
33035 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33039 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33041 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33043 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33045 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33046 Change int to HOST_WIDE_INT.
33047 * config/aarch64/aarch64-protos.h
33048 (aarch64_simd_gen_const_vector_dup): Likewise.
33049 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33051 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33053 * langhooks-def.h (lhd_type_for_size): New decl.
33054 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33055 * langhooks.c (lhd_type_for_size): New function, taken from
33058 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33060 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33061 define_bypass for CR latency.
33062 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33063 (power9-alu2): Add define_bypass for CR latency.
33065 (power9-mul): Update insn latency.
33066 (power9-mul-compare): Update insn latency, bypass latency and remove
33069 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33071 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33073 * config/aarch64/aarch64.md
33074 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33075 aarch64_nopcrelative_literal_loads.
33076 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33078 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33080 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33081 TARGET_LOONGSON_3A.
33082 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33084 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33087 * config.gcc (supported_defaults): Add lxc1-sxc1.
33088 (with_lxc1_sxc1): Add validation.
33089 (all_defaults): Add lxc1-sxc1.
33090 * config/mips/mips.opt (mlxc1-sxc1): New option.
33091 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33093 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33094 __mips_no_lxc1_sxc1.
33095 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33096 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33097 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33099 2017-01-19 Richard Biener <rguenther@suse.de>
33101 PR tree-optimization/72488
33102 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33103 sure to restore SSA info.
33104 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33106 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33108 PR rtl-optimization/79121
33109 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33110 of the inner type when shifting an extended value.
33112 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33115 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33116 interposable aliases.
33118 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33121 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33122 Use the evmergelohi instruction.
33123 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33124 (mov_sitf_e500_subreg8_2_be): Likewise.
33125 (mov_sitf_e500_subreg12_2_le): Likewise.
33126 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33127 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33128 (mov_sitf_e500_subreg8_2_le): Likewise.
33129 (mov_sitf_e500_subreg12_2_be): Likewise.
33131 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33133 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33134 attribute from vecsimple to vecperm.
33135 (altivec_vbpermq2): Likewise.
33137 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33140 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33142 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33143 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33144 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33145 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33146 case where N arg is SIZE_MAX.
33147 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33148 (cmpstrsi): Add pattern.
33150 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33152 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33153 __builtin_vec_revb builtins.
33154 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33155 built-in functions to support generation of the ISA 3.0 XXBR<x>
33156 vector byte reverse instructions.
33157 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33158 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33159 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33160 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33161 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33162 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33163 (P9V_BUILTIN_VEC_REVB): Likewise.
33164 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33165 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33166 (p9_xxbrq_v16qi): Likewise.
33167 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33168 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33169 (p9_xxbrh_v8hi): Likewise.
33170 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33171 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33172 vec_revb built-in functions.
33174 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33176 PR rtl-optimization/78952
33177 * config/i386/i386.md (any_extract): New code iterator.
33178 (*insvqi_2): Use any_extract for source operand.
33179 (*insvqi_3): Use any_shiftrt for source operand.
33181 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33183 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33185 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33187 2017-01-18 Matthias Klose <doko@ubuntu.com>
33189 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33191 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33193 * config/rs6000/altivec.h (vec_bperm): Change #define.
33194 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33195 (altivec_vbpermq2): New define_insn.
33196 (altivec_vbpermd): Likewise.
33197 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33198 function interface.
33199 (VBPERMD): Likewise.
33200 (VBPERM): New polymorphic function interface.
33201 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33202 Add entries for P9V_BUILTIN_VEC_VBPERM.
33203 * doc/extend.texi: Add interfaces for vec_bperm.
33205 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33207 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33208 first letter of error messages.
33209 (s390_resolve_overloaded_builtin): Likewise.
33210 * config/s390/s390.c (s390_expand_builtin): Likewise.
33211 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33212 (s390_valid_target_attribute_inner_p): Likewise.
33213 * config/s390/s390.md ("tabort"): Likewise.
33215 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33217 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33218 (ISA_AVOID_DIV_HILO): New macro.
33219 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33220 (ISA_HAS_DDIV): Likewise.
33222 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33224 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33226 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33228 * doc/invoke.texi (fabi-version): Spelling fix.
33230 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33233 * doc/invoke.texi (fabi-version): Mention mangling fix for
33236 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33239 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33241 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33244 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33245 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33247 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33249 (stack_protect_test): Ditto.
33250 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33251 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33253 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33254 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33255 -mstack-protector-guard-offset=.
33256 (RS/6000 and PowerPC Options): Ditto.
33258 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33260 * config/i386/i386.h (MASK_CLASS_P): New define.
33261 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33262 there are no registers from different register sets also when
33263 mask registers are used. Update function comment.
33264 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33265 to (*k/*r) and (*k/*km) alternatives.
33267 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33269 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33270 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33271 (EH_RETURN_HANDLER_RTX): New define.
33272 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33273 Force frame pointer in EH return functions.
33274 (aarch64_expand_epilogue): Add barrier for eh_return.
33275 (aarch64_final_eh_return_addr): Remove.
33276 (aarch64_eh_return_handler_rtx): New function.
33277 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33279 (aarch64_eh_return_handler_rtx): New prototype.
33281 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33283 * config/rs6000/altivec.h (vec_rlmi): New #define.
33284 (vec_vrlnm): Likewise.
33285 (vec_rlnm): Likewise.
33286 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33287 (UNSPEC_VRLNM): Likewise.
33288 (VIlong): New mode iterator.
33289 (altivec_vrl<VI_char>mi): New define_insn.
33290 (altivec_vrl<VI_char>nm): Likewise.
33291 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33293 (VRLDNM): Likewise.
33294 (RLNM): New polymorphic function entry.
33295 (VRLWMI): New monomorphic function entry.
33296 (VRLDMI): Likewise.
33297 (RLMI): New polymorphic function entry.
33298 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33299 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33300 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33303 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33306 * dwarf2out.c (field_byte_offset): Restore the
33307 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33308 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33309 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33312 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33315 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33317 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33320 * dwarf2out.c (add_data_member_location_attribute): For constant
33321 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33322 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33323 DW_AT_byte_size attributes.
33325 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33327 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33328 after forcing to constant memory when the code model is medium.
33330 2017-01-17 Julia Koval <julia.koval@intel.com>
33333 * config/i386/avx512fintrin.h
33334 (_mm512_i32gather_ps): Change __addr type to void const*.
33335 (_mm512_mask_i32gather_ps): Ditto.
33336 (_mm512_i32gather_pd): Ditto.
33337 (_mm512_mask_i32gather_pd): Ditto.
33338 (_mm512_i64gather_ps): Ditto.
33339 (_mm512_mask_i64gather_ps): Ditto.
33340 (_mm512_i64gather_pd): Ditto.
33341 (_mm512_mask_i64gather_pd): Ditto.
33342 (_mm512_i32gather_epi32): Ditto.
33343 (_mm512_mask_i32gather_epi32): Ditto.
33344 (_mm512_i32gather_epi64): Ditto.
33345 (_mm512_mask_i32gather_epi64): Ditto.
33346 (_mm512_i64gather_epi32): Ditto.
33347 (_mm512_mask_i64gather_epi32): Ditto.
33348 (_mm512_i64gather_epi64): Ditto.
33349 (_mm512_mask_i64gather_epi64): Ditto.
33350 (_mm512_i32scatter_ps): Change __addr type to void*.
33351 (_mm512_mask_i32scatter_ps): Ditto.
33352 (_mm512_i32scatter_pd): Ditto.
33353 (_mm512_mask_i32scatter_pd): Ditto.
33354 (_mm512_i64scatter_ps): Ditto.
33355 (_mm512_mask_i64scatter_ps): Ditto.
33356 (_mm512_i64scatter_pd): Ditto.
33357 (_mm512_mask_i64scatter_pd): Ditto.
33358 (_mm512_i32scatter_epi32): Ditto.
33359 (_mm512_mask_i32scatter_epi32): Ditto.
33360 (_mm512_i32scatter_epi64): Ditto.
33361 (_mm512_mask_i32scatter_epi64): Ditto.
33362 (_mm512_i64scatter_epi32): Ditto.
33363 (_mm512_mask_i64scatter_epi32): Ditto.
33364 (_mm512_i64scatter_epi64): Ditto.
33365 (_mm512_mask_i64scatter_epi64): Ditto.
33366 * config/i386/avx512pfintrin.h
33367 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33368 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33369 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33370 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33371 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33372 (_mm512_prefetch_i32scatter_ps): Ditto.
33373 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33374 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33375 (_mm512_prefetch_i64scatter_pd): Ditto.
33376 (_mm512_prefetch_i64scatter_ps): Ditto.
33377 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33378 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33379 * config/i386/avx512vlintrin.h
33380 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33381 (_mm_mmask_i32gather_ps): Ditto.
33382 (_mm256_mmask_i32gather_pd): Ditto.
33383 (_mm_mmask_i32gather_pd): Ditto.
33384 (_mm256_mmask_i64gather_ps): Ditto.
33385 (_mm_mmask_i64gather_ps): Ditto.
33386 (_mm256_mmask_i64gather_pd): Ditto.
33387 (_mm_mmask_i64gather_pd): Ditto.
33388 (_mm256_mmask_i32gather_epi32): Ditto.
33389 (_mm_mmask_i32gather_epi32): Ditto.
33390 (_mm256_mmask_i32gather_epi64): Ditto.
33391 (_mm_mmask_i32gather_epi64): Ditto.
33392 (_mm256_mmask_i64gather_epi32): Ditto.
33393 (_mm_mmask_i64gather_epi32): Ditto.
33394 (_mm256_mmask_i64gather_epi64): Ditto.
33395 (_mm_mmask_i64gather_epi64): Ditto.
33396 (_mm256_i32scatter_ps): Change __addr type to void*.
33397 (_mm256_mask_i32scatter_ps): Ditto.
33398 (_mm_i32scatter_ps): Ditto.
33399 (_mm_mask_i32scatter_ps): Ditto.
33400 (_mm256_i32scatter_pd): Ditto.
33401 (_mm256_mask_i32scatter_pd): Ditto.
33402 (_mm_i32scatter_pd): Ditto.
33403 (_mm_mask_i32scatter_pd): Ditto.
33404 (_mm256_i64scatter_ps): Ditto.
33405 (_mm256_mask_i64scatter_ps): Ditto.
33406 (_mm_i64scatter_ps): Ditto.
33407 (_mm_mask_i64scatter_ps): Ditto.
33408 (_mm256_i64scatter_pd): Ditto.
33409 (_mm256_mask_i64scatter_pd): Ditto.
33410 (_mm_i64scatter_pd): Ditto.
33411 (_mm_mask_i64scatter_pd): Ditto.
33412 (_mm256_i32scatter_epi32): Ditto.
33413 (_mm256_mask_i32scatter_epi32): Ditto.
33414 (_mm_i32scatter_epi32): Ditto.
33415 (_mm_mask_i32scatter_epi32): Ditto.
33416 (_mm256_i32scatter_epi64): Ditto.
33417 (_mm256_mask_i32scatter_epi64): Ditto.
33418 (_mm_i32scatter_epi64): Ditto.
33419 (_mm_mask_i32scatter_epi64): Ditto.
33420 (_mm256_i64scatter_epi32): Ditto.
33421 (_mm256_mask_i64scatter_epi32): Ditto.
33422 (_mm_i64scatter_epi32): Ditto.
33423 (_mm_mask_i64scatter_epi32): Ditto.
33424 (_mm256_i64scatter_epi64): Ditto.
33425 (_mm256_mask_i64scatter_epi64): Ditto.
33426 (_mm_i64scatter_epi64): Ditto.
33427 (_mm_mask_i64scatter_epi64): Ditto.
33428 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33429 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33430 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33431 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33432 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33433 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33434 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33435 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33436 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33437 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33438 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33439 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33440 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33441 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33442 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33443 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33444 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33445 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33446 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33447 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33448 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33449 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33450 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33451 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33452 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33453 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33454 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33455 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33456 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33457 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33458 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33459 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33460 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33461 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33462 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33463 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33464 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33465 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33466 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33467 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33468 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33469 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33470 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33471 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33472 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33473 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33474 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33475 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33476 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33477 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33478 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33479 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33480 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33481 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33482 definitions accordingly.
33484 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33485 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33488 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33491 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33494 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33495 hard regs for inner regno.
33497 2017-01-17 Martin Liska <mliska@suse.cz>
33500 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33501 assumption and add comment.
33503 2017-01-17 Nathan Sidwell <nathan@acm.org>
33505 * ipa-visibility.c (localize_node): New function, broken out of ...
33506 (function_and_variable_visibility): ... here. Call it.
33508 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33510 PR middle-end/77445
33511 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33512 correctly set frequency of oudgoing edge.
33513 (duplicate_thread_path): Fix profile updating.
33515 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33518 * configure.ac: Add GCC_BASE_VER.
33519 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33520 version from BASE-VER file.
33521 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33522 (gcc.o): Depend on $(BASEVER).
33523 * common.opt (dumpfullversion): New option.
33524 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33525 * doc/invoke.texi: Document -dumpfullversion.
33526 * doc/install.texi: Document --with-gcc-major-version-only.
33527 * configure: Regenerated.
33529 2017-01-17 Richard Biener <rguenther@suse.de>
33531 PR tree-optimization/71433
33532 * tree-vrp.c (register_new_assert_for): Merge same asserts
33533 on all incoming edges.
33534 (process_assert_insertions_for): Handle insertions at the
33537 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33539 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33540 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33542 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33545 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33548 2017-01-17 Alan Modra <amodra@gmail.com>
33551 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33552 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33553 symbolic stack limit when pic.
33555 2017-01-16 Martin Sebor <msebor@redhat.com>
33557 PR tree-optimization/78608
33558 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
33560 2017-01-16 Jeff Law <law@redhat.com>
33563 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
33564 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
33565 for several include directories that may be relative to sysroot.
33566 * config/i386/x-mingw32 (gplus_includedir): Define.
33567 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
33568 (native_system_includedir): Likewise.
33569 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
33570 override if TARGET_SYSTEM_ROOT is defined.
33571 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
33573 PR tree-optimization/79090
33574 PR tree-optimization/33562
33575 PR tree-optimization/61912
33576 PR tree-optimization/77485
33577 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
33578 and computed trims into the dump file.
33580 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33582 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
33584 2017-01-16 Jakub Jelinek <jakub@redhat.com>
33587 * gimplify.c (gimplify_init_constructor): If want_value and
33588 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
33592 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
33593 sequence. Formatting fixes.
33594 (doloop_optimize): Formatting fixes.
33597 * gcc.c (debug_level_greater_than_spec_func): New function.
33598 (static_spec_functions): Add debug-level-gt spec function.
33599 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
33601 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33602 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33603 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
33604 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
33605 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
33606 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
33608 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
33610 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
33611 QImode fixups to general and mask registers only.
33613 2017-01-16 Carl Love <cel@us.ibm.com>
33615 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
33616 for built-in functions
33617 vector signed char vec_nabs (vector signed char)
33618 vector signed short vec_nabs (vector signed short)
33619 vector signed int vec_nabs (vector signed int)
33620 vector signed long long vec_nabs (vector signed long long)
33621 vector float vec_nabs (vector float)
33622 vector double vec_nabs (vector double)
33623 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
33625 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
33626 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
33627 * doc/extend.texi: Update the documentation file for the new built-in
33630 2017-01-16 Martin Sebor <msebor@redhat.com>
33632 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
33635 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33637 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
33638 UNSPEC_VSX__XXSPLTD to require special splat handling.
33640 2017-01-16 David Malcolm <dmalcolm@redhat.com>
33643 * system.h: Poison strndup.
33645 2017-01-16 Alan Modra <amodra@gmail.com>
33648 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
33651 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
33653 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
33655 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
33657 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
33658 call recog here. Assert that INSN_CODE (insn) is non-negative.
33660 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33663 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
33665 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
33666 in the currently scheduled RTL fragment.
33668 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33670 PR rtl-optimization/78751
33671 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
33674 2017-01-14 Jeff Law <law@redhat.com>
33676 PR tree-optimization/79090
33677 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
33678 variable length stores.
33679 (compute_trims): Delete dead assignment to *trim_tail.
33680 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
33683 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
33685 PR rtl-optimization/78626
33686 PR rtl-optimization/78727
33687 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
33688 of a block, and split such blocks after everything else is finished.
33690 2017-01-14 Alan Modra <amodra@gmail.com>
33693 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
33694 target legitimate_combined_insn.
33695 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
33696 (rs6000_legitimate_combined_insn): New function.
33697 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
33699 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
33700 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
33701 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
33703 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
33705 * doc/frontends.texi (G++ and GCC): Remove references to Java.
33707 2017-01-13 Jeff Law <law@redhat.com>
33709 PR tree-optimization/33562
33710 PR tree-optimization/61912
33711 PR tree-optimization/77485
33712 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
33714 (delete_dead_assignment): Likewise.
33715 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
33716 statement to delete_dead_call and delete_dead_assignment.
33718 2017-01-13 David Malcolm <dmalcolm@redhat.com>
33721 * substring-locations.c (format_warning_va): Strengthen case 1 so
33722 that both endpoints of the substring must be within the format
33723 range for just the substring to be printed.
33725 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
33727 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
33728 * config/i386/i386.c (ix86_target_string): Add missing options
33729 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
33730 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
33731 flags_other and ix86_target_other to flags2_other. Display unknown
33733 (ix86_valid_target_attribute_inner_p): Add missing options and
33734 reorder options by implied ISAs, as in ix86_target_string.
33736 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33738 * hash-table.h (hash_table::too_empty_p): New function.
33739 (hash_table::expand): Use it.
33740 (hash_table::traverse): Likewise.
33741 (hash_table::empty_slot): Use sizeof (value_type) instead of
33742 sizeof (PTR) to convert bytes to elements. Shrink the table
33743 if the current size is excessive for the current number of
33746 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33748 * ira-costs.c (record_reg_classes): Break from the inner loop
33749 early once alt_fail is known to be true. Update outer loop
33750 handling accordingly.
33752 2017-01-13 Jeff Law <law@redhat.com>
33754 * tree-ssa-dse.c (decrement_count): New function.
33755 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33756 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33757 when we know the partially dead statement is a mem* function.
33759 PR tree-optimization/61912
33760 PR tree-optimization/77485
33761 * tree-ssa-dse.c: Include expr.h.
33762 (maybe_trim_constructor_store): New function.
33763 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33765 PR tree-optimization/33562
33766 PR tree-optimization/61912
33767 PR tree-optimization/77485
33768 * doc/invoke.texi: Document new dse-max-object-size param.
33769 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
33770 * tree-ssa-dse.c: Include params.h.
33771 (dse_store_status): New enum.
33772 (initialize_ao_ref_for_dse): New, partially extracted from
33774 (valid_ao_ref_for_dse, normalize_ref): New.
33775 (setup_live_bytes_from_ref, compute_trims): Likewise.
33776 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
33777 (maybe_trim_partially_dead_store): Likewise.
33778 (maybe_trim_complex_store): Likewise.
33779 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
33780 Track what bytes live from the original store. Return tri-state
33781 for dead, partially dead or live.
33782 (dse_dom_walker): Add constructor, destructor and new private members.
33783 (delete_dead_call, delete_dead_assignment): New extracted from
33785 (dse_optimize_stmt): Make a member of dse_dom_walker.
33786 Use initialize_ao_ref_for_dse.
33788 PR tree-optimization/33562
33789 PR tree-optimization/61912
33790 PR tree-optimization/77485
33791 * sbitmap.h (bitmap_count_bits): Prototype.
33792 (bitmap_clear_range, bitmap_set_range): Likewise.
33793 * sbitmap.c (bitmap_clear_range): New function.
33794 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
33796 2017-01-13 Martin Liska <mliska@suse.cz>
33799 * function.c (set_cfun): Add new argument force.
33800 * function.h (set_cfun): Likewise.
33801 * ipa-inline-transform.c (inline_call): Use the function when
33802 strict alising from is dropped for function we inline to.
33804 2017-01-13 Richard Biener <rguenther@suse.de>
33806 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
33807 for dumping GIMPLE INTEGER_CSTs.
33809 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33811 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
33812 to 201112L since C++17.
33814 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
33817 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
33818 if -fsanitize=kernel-address is present.
33820 2017-01-13 Richard Biener <rguenther@suse.de>
33822 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33823 as _Literal ( type ) number in case usual suffixes do not
33824 preserve all information.
33826 2017-01-13 Richard Biener <rguenther@suse.de>
33828 PR tree-optimization/77283
33829 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33830 and ssa-iterators.h.
33831 (is_feasible_trace): Implement a cost model based on joiner
33834 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33837 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33838 char or short to __float128/_Float128 directly.
33840 2017-01-12 Martin Sebor <msebor@redhat.com>
33842 to -Wformat-overflow.
33843 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33844 (min_bytes_remaining): Same.
33845 (get_string_length): Same.
33846 (format_string): Same.
33847 (format_directive): Same.
33849 (pass_sprintf_length::handle_gimple_call): Same.
33851 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33853 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33854 info.nowrite calls with no lhs that can't throw. Return bool
33855 whether gsi_remove has been called or not.
33856 (pass_sprintf_length::handle_gimple_call): Return bool whether
33857 try_substitute_return_value called gsi_remove. Formatting fix.
33858 (pass_sprintf_length::execute): Don't use gsi_remove if
33859 handle_gimple_call returned true.
33862 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33863 be removed due to side-effects, don't remove following barrier nor
33864 turn the successor edge into fallthru edge.
33866 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33869 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33870 element-reversing loads and stores as not swappable.
33872 2017-01-12 Nathan Sidwell <nathan@acm.org>
33873 Nicolai Stange <nicstange@gmail.com>
33875 * combine.c (try_combine): Don't ignore result of overlap checking
33876 loop. Combine overlap & asm check into single loop.
33878 2017-01-12 Richard Biener <rguenther@suse.de>
33880 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33881 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33883 2017-01-12 Richard Biener <rguenther@suse.de>
33885 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33886 and TS_TARGET_OPTION directly derive from TS_BASE.
33887 * tree-core.h (tree_optimization_option): Derive from tree_base.
33888 (tree_target_option): Likewise.
33890 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33892 * config/i386/i386.c (memory_address_length): Increase len
33893 only when rip_relative_addr_p returns false.
33895 2017-01-11 Julia Koval <julia.koval@intel.com>
33897 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33898 (OPTION_MASK_ISA_SGX_SET): New.
33899 (ix86_handle_option): Handle OPT_msgx.
33900 * config.gcc: Added sgxintrin.h.
33901 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33902 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33903 * config/i386/i386.c (ix86_target_string): Add -msgx.
33905 (ix86_option_override_internal): Handle new options.
33906 (ix86_valid_target_attribute_inner_p): Add sgx.
33907 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33908 * config/i386/i386.opt: Add msgx.
33909 * config/i386/sgxintrin.h: New file.
33910 * config/i386/x86intrin.h: Add sgxintrin.h.
33912 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33915 * fold-const.c (maybe_nonzero_address): Return 1 for function
33917 (tree_single_nonzero_warnv_p): Don't handle function local objects
33921 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33924 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33927 * opts.c: Include "spellcheck.h"
33928 (struct string_fragment): New struct.
33929 (struct edit_distance_traits<const string_fragment &>): New
33931 (get_closest_sanitizer_option): New function.
33932 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33934 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33936 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33938 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33939 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33940 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33941 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33942 for initial die_offset if dwarf_split_debug_info.
33943 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33944 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33946 (output_skeleton_debug_sections): Formatting fix. Use
33947 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33948 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33950 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33952 * config/arm/cortex-a53.md: Add bypasses for
33953 cortex_a53_r2f_cvt.
33954 (cortex_a53_r2f): Only use for transfers.
33955 (cortex_a53_f2r): Likewise.
33956 (cortex_a53_r2f_cvt): Add reservation for conversions.
33957 (cortex_a53_f2r_cvt): Likewise.
33959 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33961 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33962 to all inlined functions, change static to extern.
33964 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33967 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33969 (arm_assemble_integer): Likewise.
33971 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33973 * config.gcc: Use new awk script to check CPU, FPU and architecture
33974 parameters for --with-... options.
33975 * config/arm/parsecpu.awk: New file
33976 * config/arm/arm-cpus.in: New file.
33977 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33979 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33981 * config/arm/t-arm: Update dependency rules.
33982 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33983 of processing .def files.
33984 * config/arm/genopt.sh: Deleted.
33985 * config/arm/gentune.sh: Deleted.
33986 * config/arm/arm-cores.def: Deleted.
33987 * config/arm/arm-arches.def: Deleted.
33988 * config/arm/arm-fpus.def: Deleted.
33989 * config/arm/arm-tune.md: Regenerated.
33990 * config/arm/arm-tables.opt: Regenerated.
33991 * config/arm/arm-cpu.h: New generated file.
33992 * config/arm/arm-cpu-data.h: New generated file.
33993 * config/arm/arm-cpu-cdata.h: New generated file.
33995 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33998 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34000 (input_varpool_node): Unpack dynamically_initialized bit.
34002 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34004 PR rtl-optimization/79032
34005 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34006 the alignment of the adjusted memory reference against that of MODE,
34007 instead of the alignment of the original memory reference.
34009 2017-01-11 Martin Jambor <mjambor@suse.cz>
34011 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34013 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34014 decorated functions.
34016 2017-01-11 Richard Biener <rguenther@suse.de>
34018 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34019 set range/nonnull info for PHI results. Do not set it on
34020 stmts marked for removal.
34022 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34024 * expr.c (store_field): In the bitfield case, fetch the return value
34025 from the registers before applying a single big-endian adjustment.
34026 Always do a final load for a BLKmode value not larger than a word.
34028 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34031 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34032 that we correctly handle column numbers greater than
34033 LINE_MAP_MAX_COLUMN_NUMBER.
34035 2017-01-10 Martin Sebor <msebor@redhat.com>
34037 PR middle-end/78245
34038 * gimple-ssa-sprintf.c (get_destination_size): Call
34039 {init,fini}object_sizes.
34040 * tree-object-size.c (addr_object_size): Adjust.
34041 (pass_through_call): Adjust.
34042 (pass_object_sizes::execute): Adjust.
34043 * tree-object-size.h (fini_object_sizes): Declare.
34045 2017-01-10 Martin Sebor <msebor@redhat.com>
34047 PR tree-optimization/78775
34048 * builtins.c (get_size_range): Move...
34049 * calls.c: ...to here.
34050 (alloc_max_size): Accept zero argument.
34051 (operand_signed_p): Remove.
34052 (maybe_warn_alloc_args_overflow): Call get_size_range.
34053 * calls.h (get_size_range): Declare.
34055 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34057 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34058 from TI's devices.csv file as of September 2016.
34059 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34061 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34063 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34064 * doc/invoke.texi: Likewise.
34065 * doc/md.texi: Likewise.
34066 * doc/objc.texi: Likewise.
34068 2017-01-10 Joshua Conner <joshconner@google.com>
34070 * config/arm/fuchsia-elf.h: New file.
34071 * config/fuchsia.h: New file.
34072 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34073 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34075 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34077 2016-01-10 Richard Biener <rguenther@suse.de>
34079 PR tree-optimization/79034
34080 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34081 Propagate out degenerate PHIs in the joiner.
34083 2017-01-10 Martin Liska <mliska@suse.cz>
34085 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34086 (sort_congruence_classes_by_decl_uid): Likewise.
34087 (sort_congruence_class_groups_by_decl_uid): Likewise.
34088 (sem_item_optimizer::merge_classes): Sort class, groups in these
34089 classes and members in the groups by DECL_UID of declarations.
34090 This would make merge operations stable.
34092 2017-01-10 Martin Liska <mliska@suse.cz>
34094 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34095 usage of m_classes_vec.
34096 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34097 (sem_item_optimizer::get_group_by_hash): Likewise.
34098 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34099 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34100 (sem_item_optimizer::verify_classes): Likewise.
34101 (sem_item_optimizer::process_cong_reduction): Likewise.
34102 (sem_item_optimizer::dump_cong_classes): Likewise.
34103 (sem_item_optimizer::merge_classes): Likewise.
34104 * ipa-icf.h (congruence_class_hash): Rename from
34105 congruence_class_group_hash. Remove declaration of m_classes_vec.
34107 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34109 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34110 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34111 * config.gcc: Add avx512vpopcntdqintrin.h.
34112 * config/i386/avx512vpopcntdqintrin.h: New.
34113 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34114 * config/i386/i386-builtin-types.def: Add new types.
34115 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34116 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34117 __builtin_ia32_vpopcountq_v8di_mask): New.
34118 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34119 __AVX512VPOPCNTDQ__.
34120 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34121 (PTA_AVX512VPOPCNTDQ): Define.
34122 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34123 TARGET_AVX512VPOPCNTDQ_P): Define.
34124 * config/i386/i386.opt: Add mavx512vpopcntdq.
34125 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34126 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34128 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34130 PR middle-end/77484
34131 * predict.def (PRED_CALL): Set to 67.
34133 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34135 * expr.c (store_field): In the bitfield case, if the value comes from
34136 a function call and is of an aggregate type returned in registers, do
34137 not modify the field mode; extract the value in all cases if the mode
34138 is BLKmode and the size is not larger than a word.
34140 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34143 * config/i386/cpuid.h: Fix undefined behavior.
34145 2017-01-04 Jeff Law <law@redhat.com>
34147 PR tree-optimization/79007
34148 PR tree-optimization/67955
34149 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34150 conservative for pt.null when flag_non_call_exceptions is on.
34152 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34154 PR translation/79019
34155 PR translation/79020
34156 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34157 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34158 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34160 * config/avr/avr.opt (maccumulate-args): Likewise.
34161 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34162 * common.opt (freport-bug): Likewise.
34163 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34164 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34165 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34166 translatable string.
34167 * config/i386/i386.c (function_value_32): Likewise.
34168 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34169 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34171 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34172 * common/config/msp430/msp430-common.c (msp430_handle_option):
34174 * symtab.c (symtab_node::verify_base): Likewise.
34175 * opts.c (set_debug_level): Likewise.
34176 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34177 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34178 missing whitespace to translatable strings.
34179 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34180 * config/sh/superh.h: Likewise.
34181 * config/i386/xopintrin.h: Likewise.
34182 * config/i386/znver1.md: Likewise.
34183 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34184 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34185 * double-int.h (struct double_int): Likewise.
34186 * double-int.c (div_and_round_double): Likewise.
34187 * wide-int.cc: Likewise.
34188 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34189 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34190 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34191 (crossjumps_occurred): ... this.
34192 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34195 PR tree-optimization/78899
34196 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34197 returning bool return struct loop *, NULL for failure and the new
34199 (versionable_outer_loop_p): Don't version outer loop if it has
34200 dont_vectorized bit set.
34201 (tree_if_conversion): When versioning outer loop, ensure
34202 tree_if_conversion is performed also on the inner loop of the
34203 non-vectorizable outer loop copy.
34204 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34205 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34206 prevent vectorization of it.
34207 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34208 the outer loop vectorization of the non-scalar version is attempted
34209 before vectorization of the inner loop in scalar version. If
34210 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34211 vectorization of its inner loop.
34212 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34213 has 2 inner loops, rename also on edges from bb whose single pred
34214 is outer_loop->header. Fix typo in function comment.
34216 2017-01-09 Martin Sebor <msebor@redhat.com>
34219 * asan.c (asan_emit_stack_protection): Increase local buffer size
34220 to avoid snprintf truncation warning.
34222 2017-01-09 Andrew Pinski <apinski@cavium.com>
34224 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34225 to reference thunderx2t99 for the tuning structure
34226 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34228 (thunderx2t99_extra_costs): This.
34229 * config/aarch64/aarch64-tune.md: Regenerate.
34230 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34231 (vulcan_addrcost_table): This.
34232 (vulcan_regmove_cost): Rename to ...
34233 (thunderx2t99_regmove_cost): This.
34234 (vulcan_vector_cost): Rename to ...
34235 (thunderx2t99_vector_cost): this.
34236 (vulcan_branch_cost): Rename to ...
34237 (thunderx2t99_branch_cost): This.
34238 (vulcan_tunings): Rename to ...
34239 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34240 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34242 2017-01-09 Martin Jambor <mjambor@suse.cz>
34246 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34247 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34248 (propagate_vr_accross_jump_function): Use the above function for all
34249 value range computations for pass-through jump functions and type
34250 converasion from explicit value range values.
34251 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34252 parameters from TYPE_ARG_TYPES.
34253 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34254 (ipa_write_node_info): Stream type of the actual argument.
34255 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34257 2017-01-09 Martin Liska <mliska@suse.cz>
34260 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34261 (lookup_compiler): Do not show error message with have_E.
34263 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34265 PR tree-optimization/78938
34266 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34267 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34268 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34269 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34272 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34274 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34277 2017-01-09 Richard Biener <rguenther@suse.de>
34279 PR tree-optimization/78997
34280 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34281 name condition properly.
34283 2017-01-09 Richard Biener <rguenther@suse.de>
34286 * dwarf2out.c (is_cxx): New overload with context.
34287 (is_naming_typedef_decl): Use it.
34289 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34291 * invoke.texi (Option Summary): Correct spacing in option lists
34292 and add line breaks to fix over-long lines.
34294 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34296 PR middle-end/17660
34298 * extend.texi (Common Variable Attributes): Add xref to GCC
34299 Internals manual to explain mode attribute keywords.
34301 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34304 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34305 and Preprocessor Options.
34306 (Options for Linking): Document -pthread here....
34307 (RS/6000 and PowerPC Options): ...not here.
34308 (Solaris 2 Options): ...or here.
34309 * doc/cppopts.texi: Document -pthread.
34311 2017-01-08 Martin Sebor <msebor@redhat.com>
34313 PR middle-end/77708
34314 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34315 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34316 New member functions.
34317 (format_directive): Used them.
34319 (pass_sprintf_length::handle_gimple_call): Same.
34320 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34321 to avoid truncation for any argument.
34322 (extract_affine_mul): Same.
34323 * tree.c (get_file_function_name): Same.
34325 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34327 PR middle-end/77484
34328 * predict.def (PRED_INDIR_CALL): Set to 86.
34330 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34332 PR preprocessor/54124
34333 * doc/cppopts.texi: Reformat -d subtable to list the full name
34334 of the options. Add cross-reference to the docs for the general
34335 compiler -d options.
34336 * doc/invoke.texi (Developer Options): Add cross-reference to the
34337 preprocessor-specific -d option documentation.
34339 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34341 PR preprocessor/13498
34342 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34343 redudant material, and reflect new command-line options.
34344 (System Headers): Likewise.
34346 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34348 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34349 -isystem, and -idirafter. Copy-edit.
34350 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34351 default for -ftrack-macro-expansion. Delete obsolete and
34352 badly-formatted implementation details about -fdebug-cpp output.
34353 * doc/cppwarnopts.texi: Copy-edit.
34355 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34358 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34359 that the transition from a max line width >= 1<<10 to narrower
34360 lines works correctly.
34362 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34364 * doc/options.texi (PerFunction): New.
34365 * opt-functions.awk (switch_flags): Map both Optimization and
34366 PerFunction to CL_OPTIMIZATION.
34367 * opth-gen.awk: Test for PerFunction flag along with
34369 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34370 it only when the latter is present. Skip those that don't in
34371 the hash function generator.
34372 * common.opt (fvar-tracking): Mark as PerFunction instead of
34374 (fvar-tracking-assignments): Likewise.
34375 (fvar-tracking-assignments-toggle): Likewise.
34376 (fvar-tracking-uninit): Likewise.
34378 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34380 PR translation/79018
34381 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34384 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34387 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34388 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34389 TARGET_LONG_JUMP_TABLE_OFFSETS.
34390 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34391 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34392 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34393 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34394 * config/m68k/m68k.md (tablejump expander): Likewise.
34395 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34396 TARGET_LONG_JUMP_TABLE_OFFSETS.
34397 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34398 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34400 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34401 David Holsgrove <david.holsgrove@xilinx.com>
34403 * common/config/microblaze/microblaze-common.c
34404 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34405 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34407 * config/microblaze/microblaze.c (microblaze_must_save_register)
34408 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34410 (microblaze_eh_return): New function.
34411 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34412 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34413 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34414 * config/microblaze/microblaze.md (eh_return): New pattern.
34416 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34418 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34419 GCC_DIAGNOSTIC_STRINGIFY): Define.
34421 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34423 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34425 * config/arm/arm.md (<mcrr>): New.
34427 * config/arm/arm.c (arm_arch5te): New.
34428 (arm_option_override): Set arm_arch5te.
34429 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34431 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34432 (arm_mcrr_qualifiers): ... this. New.
34433 (MRRC_QUALIFIERS): Define to...
34434 (arm_mrrc_qualifiers): ... this. New.
34435 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34437 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34438 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34439 (MRRCI, mrrc, MRRC): New.
34440 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34441 VUNSPEC_MRRC2): New.
34443 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34445 * config/arm/arm.md (<mcr>): New.
34447 * config/arm/arm.c (arm_coproc_builtin_available): Add
34448 support for mcr, mrc, mcr2 and mrc2.
34449 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34450 (arm_mcr_qualifiers): ... this. New.
34451 (MRC_QUALIFIERS): Define to ...
34452 (arm_mrc_qualifiers): ... this. New.
34453 (MCR_QUALIFIERS): Define to ...
34454 (arm_mcr_qualifiers): ... this. New.
34455 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34457 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34458 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34459 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34460 VUNSPEC_MRC2): New.
34462 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34464 * config/arm/arm.md (*ldc): New.
34468 * config/arm/arm.c (arm_coproc_builtin_available): Add
34469 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34470 (arm_coproc_ldc_stc_legitimate_address): New.
34471 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34472 'qualifier_const_pointer'.
34473 (LDC_QUALIFIERS): Define to...
34474 (arm_ldc_qualifiers): ... this. New.
34475 (STC_QUALIFIERS): Define to...
34476 (arm_stc_qualifiers): ... this. New.
34477 * config/arm/arm-protos.h
34478 (arm_coproc_ldc_stc_legitimate_address): New.
34479 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34480 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34481 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34482 stc2, stcl, stc2l): New.
34483 * config/arm/constraints.md (Uz): New.
34484 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34485 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34486 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34487 VUNSPEC_STC2L): New.
34489 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34491 * config/arm/arm.md (<cdp>): New.
34492 * config/arm/arm.c (neon_const_bounds): Rename this ...
34493 (arm_const_bounds): ... this.
34494 (arm_coproc_builtin_available): New.
34495 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34496 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34497 (CDP_QUALIFIERS): Define to...
34498 (arm_cdp_qualifiers): ... this. New.
34500 (arm_expand_builtin_args): Add case for 6 arguments.
34501 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34502 (arm_const_bounds): ... this.
34503 (arm_coproc_builtin_available): New.
34504 * config/arm/arm_acle.h (__arm_cdp): New.
34506 * config/arm/arm_acle_builtins.def (cdp): New.
34508 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34509 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34510 'arm_const_bounds'.
34511 * config/arm/types.md (coproc): New.
34512 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34513 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34514 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34515 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34517 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34519 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34520 (UBINOP_QUALIFIERS): New.
34522 (acle_builtin_data): New. Change comment.
34523 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34524 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34525 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34526 arm_acle_builtins.def.
34527 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34528 (arm_init_acle_builtins): New.
34529 (CRC32_BUILTIN): Remove.
34530 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34531 crc32cb, crc32ch and crc32cw.
34532 (arm_init_crc32_builtins): Remove.
34533 (arm_init_builtins): Use arm_init_acle_builtins rather
34534 than arm_init_crc32_builtins.
34535 (arm_expand_acle_builtin): New.
34536 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34537 * config/arm/arm_acle_builtins.def: New.
34539 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34541 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34542 (arm_builtin_datum): ... this.
34543 (arm_init_neon_builtin): Rename to ...
34544 (arm_init_builtin): ... this. Add a new parameters PREFIX
34545 and USE_SIG_IN_NAME.
34546 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34547 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34548 'arm_builtin_datum'.
34549 (arm_init_vfp_builtins): Likewise.
34550 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34551 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34552 (arm_expand_neon_args): Rename to ...
34553 (arm_expand_builtin_args): ... this. Rename builtin_arg
34554 enum values and differentiate between ARG_BUILTIN_MEMORY
34555 and ARG_BUILTIN_NEON_MEMORY.
34556 (arm_expand_neon_builtin_1): Rename to ...
34557 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
34558 values, arm_expand_builtin_args and add bool parameter NEON.
34559 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
34560 (arm_expand_vfp_builtin): Likewise.
34561 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
34563 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34565 PR middle-end/77484
34566 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
34567 * predict.c (tree_estimate_probability_bb): Reverse direction of
34568 polymorphic call predictor.
34570 2017-01-06 David Malcolm <dmalcolm@redhat.com>
34572 * passes.c (execute_one_pass): Split out pass-skipping logic into...
34573 (determine_pass_name_match): ...this new function and...
34574 (should_skip_pass_p): ...this new function.
34576 2017-01-06 Nathan Sidwell <nathan@acm.org>
34578 * ipa-visibility.c (function_and_variable_visibility): Reformat
34579 comments and long lines. Remove extrneous if.
34580 * symtab.c (symtab_node::make_decl_local): Fix code format.
34581 (symtab_node::set_section_for_node): Fix comment typo.
34583 2017-01-06 Martin Liska <mliska@suse.cz>
34586 * lra-constraints.c: Rename invariant to lra_invariant.
34587 * predict.c (set_even_probabilities): Initialize e to NULL.
34589 2017-01-05 Martin Sebor <msebor@redhat.com>
34591 PR tree-optimization/78910
34592 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
34593 (format_integer): Correct off-by-one error in the handling
34594 of precision with negative numbers in signed conversions..
34596 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
34598 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
34600 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34602 PR tree-optimization/71016
34603 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
34604 factor_out_conditional_conversion. Formatting fix.
34605 (factor_out_conditional_conversion): Add cond_stmt argument.
34606 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
34607 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
34610 2017-01-05 David Malcolm <dmalcolm@redhat.com>
34612 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
34613 read-rtl-function.o, and selftest-rtl.o.
34614 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
34615 (selftest::aarch64_test_loading_full_dump): New function.
34616 (selftest::aarch64_run_selftests): New function.
34617 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
34618 selftest::aarch64_run_selftests.
34619 * config/i386/i386.c
34620 (selftest::ix86_test_loading_dump_fragment_1): New function.
34621 (selftest::ix86_test_loading_call_insn): New function.
34622 (selftest::ix86_test_loading_full_dump): New function.
34623 (selftest::ix86_test_loading_unspec): New function.
34624 (selftest::ix86_run_selftests): Call the new functions.
34625 * emit-rtl.c (maybe_set_max_label_num): New function.
34626 * emit-rtl.h (maybe_set_max_label_num): New decl.
34627 * function.c (instantiate_decls): Guard call to
34628 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
34629 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
34631 * gensupport.c (gen_reader::gen_reader): Pass "false"
34632 for new "compact" param of rtx_reader.
34633 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
34634 rather than an empty string for NULL strings.
34635 * read-md.c: Potentially include config.h rather than bconfig.h.
34636 Wrap include of errors.h with #ifdef GENERATOR_FILE.
34637 (have_error): New global, copied from errors.c.
34638 (md_reader::read_name): Rename to...
34639 (md_reader::read_name_1): ...this, adding "out_loc" param,
34640 and converting "missing name or number" to returning false, rather
34642 (md_reader::read_name): Reimplement in terms of read_name_1.
34643 (md_reader::read_name_or_nil): New function.
34644 (md_reader::read_string): Handle "(nil)" by returning NULL.
34645 (md_reader::md_reader): Add new param "compact".
34646 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
34647 (md_reader::read_file): New method.
34648 * read-md.h (md_reader::md_reader): Add new param "compact".
34649 (md_reader::read_file): New method.
34650 (md_reader::is_compact): New accessor.
34651 (md_reader::read_name): Convert return type from void to file_location.
34652 (md_reader::read_name_or_nil): New decl.
34653 (md_reader::read_name_1): New decl.
34654 (md_reader::m_compact): New field.
34655 (noop_reader::noop_reader): Pass "false" for new "compact" param
34657 (rtx_reader::rtx_reader): Add new "compact" param.
34658 (rtx_reader::read_rtx_operand): Make virtual and convert return
34659 type from void to rtx.
34660 (rtx_reader::read_until): New decl.
34661 (rtx_reader::handle_any_trailing_information): New virtual function.
34662 (rtx_reader::postprocess): New virtual function.
34663 (rtx_reader::finalize_string): New virtual function.
34664 (rtx_reader::m_in_call_function_usage): New field.
34665 (rtx_reader::m_reuse_rtx_by_id): New field.
34666 * read-rtl-function.c: New file.
34667 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
34668 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
34669 (selftest::verify_three_block_rtl_cfg): New decl.
34670 * read-rtl-function.h: New file.
34671 * read-rtl.c: Potentially include config.h rather than bconfig.h.
34672 For host, include function.h, memmodel.h, and emit-rtl.h.
34673 (one_time_initialization): New function.
34674 (struct compact_insn_name): New struct.
34675 (compact_insn_names): New array.
34676 (find_code): Handle insn codes in compact dumps.
34677 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
34678 (bind_subst_iter_and_attr): Likewise.
34679 (add_condition_to_string): Likewise.
34680 (add_condition_to_rtx): Likewise.
34681 (apply_attribute_uses): Likewise.
34682 (add_current_iterators): Likewise.
34683 (apply_iterators): Likewise.
34684 (initialize_iterators): Guard usage of apply_subst_iterator with
34685 #ifdef GENERATOR_FILE.
34686 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
34687 (md_reader::read_mapping): Likewise.
34688 (add_define_attr_for_define_subst): Likewise.
34689 (add_define_subst_attr): Likewise.
34690 (read_subst_mapping): Likewise.
34691 (check_code_iterator): Likewise.
34692 (rtx_reader::read_rtx): Likewise. Move one-time initialization
34694 (one_time_initialization): New function.
34695 (rtx_reader::read_until): New method.
34696 (read_flags): New function.
34697 (parse_reg_note_name): New function.
34698 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
34699 Handle reuse_rtx ids.
34700 Wrap iterator lookup within #ifdef GENERATOR_FILE.
34701 Add parsing support for RTL dumps, mirroring the special-cases in
34702 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
34703 values, and calling handle_any_trailing_information.
34704 (rtx_reader::read_rtx_operand): Convert return type from void
34705 to rtx, returning return_rtx. Handle case 'e'. Call
34706 finalize_string on XSTR and XTMPL fields.
34707 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
34708 "(nil)" values were omitted. Call the postprocess vfunc on the
34710 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
34711 class ctor. Initialize m_in_call_function_usage. Call
34712 one_time_initialization.
34713 * rtl-tests.c (selftest::test_uncond_jump): Call
34714 set_new_first_and_last_insn.
34715 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
34716 * selftest-rtl.c: New file.
34717 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
34718 (selftest::get_insn_by_uid): New decl.
34719 * selftest-run-tests.c (selftest::run_tests): Call
34720 read_rtl_function_c_tests.
34721 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
34722 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
34725 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
34727 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
34728 operands in a special way. Assert that pos+len <= mode precision.
34730 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34732 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
34733 3 argument Alias with unlimited for the negative form.
34734 (fno-vect-cost-model): Removed.
34736 2017-01-05 Martin Liska <mliska@suse.cz>
34738 * hsa-gen.c (gen_hsa_divmod): New function.
34739 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34741 2017-01-05 Martin Liska <mliska@suse.cz>
34744 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34747 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34749 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34750 small constant length operands.
34752 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34754 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34755 between loop iterations.
34757 2017-01-05 Martin Liska <mliska@suse.cz>
34760 * gimplify.c (gimplify_decl_expr): Compare to
34761 asan_poisoned_variables instread of checking flags.
34762 (gimplify_target_expr): Likewise.
34763 (gimplify_expr): Likewise.
34764 (gimplify_function_tree): Conditionally initialize
34765 asan_poisoned_variables.
34767 2017-01-04 Jeff Law <law@redhat.com>
34769 PR tree-optimizatin/78812
34770 * rtl.h (contains_mem_rtx_p): Prototype.
34771 * ifcvt.c (containts_mem_rtx_p): Move from here to...
34772 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
34773 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
34774 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
34775 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
34777 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34779 * input.c (assert_char_at_range): Default-initialize actual_range.
34781 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34783 * df-scan.c (df_ref_create_structure): Make regno unsigned,
34784 to match the caller.
34786 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34788 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
34789 insns after final jump in test to emit dummy move.
34791 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34793 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
34794 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
34796 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34798 * multiple_target.c (create_dispatcher_calls): Init e_next.
34799 * tree-ssa-loop-split.c (split_loop): Init border.
34800 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
34803 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
34808 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
34809 (altivec_register_operand): Do not return true if the operand
34810 contains a SUBREG mixing SImode and SFmode.
34811 (vsx_register_operand): Likewise.
34812 (vsx_reg_sfsubreg_ok): New predicate.
34813 (vfloat_operand): Do not return true if the operand contains a
34814 SUBREG mixing SImode and SFmode.
34815 (vint_operand): Likewise.
34816 (vlogical_operand): Likewise.
34817 (gpc_reg_operand): Likewise.
34818 (int_reg_operand): Likewise.
34819 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34820 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34821 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34823 (rs6000_emit_move_si_sf_subreg): New helper function.
34824 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34825 fixup SUBREGs involving SImode and SFmode.
34826 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34827 numbers for the new peephole2 optimization.
34828 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34829 the GLIBC math library that do AND/IOR/XOR operations on single
34830 precision floating point.
34831 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34832 target macros to say whether we need to avoid SUBREGs mixing
34834 (TARGET_ALLOW_SF_SUBREG): Likewise.
34835 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34836 (UNSPEC_SI_FROM_SF): Likewise.
34837 (iorxor): Change spacing.
34838 (and_ior_xor): New iterator for AND, IOR, and XOR.
34839 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34840 (movdi_from_sf_zero_ext): Likewise.
34841 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34842 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34843 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34844 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34845 (fms<mode>4): Likewise.
34846 (fnma<mode>4): Likewise.
34847 (fnms<mode>4): Likewise.
34848 (nfma<mode>4): Likewise.
34849 (nfms<mode>4): Likewise.
34851 2017-01-04 Marek Polacek <polacek@redhat.com>
34854 * doc/invoke.texi: Document -Wpointer-compare.
34856 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34858 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34861 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34862 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34865 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34868 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34869 documentation of the powerpc_popcntb_ok attribute.
34870 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34871 code to issue warning messages if a requested CPU configuration is
34872 not supported by the binary (assembler and loader) toolchain.
34873 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34874 made to define a built-in function that has been disabled.
34875 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34876 made to define a built-in function that has been disabled.
34877 (altivec_init_builtins): Add comment explaining why definition
34878 of the DST built-in functions is not preceded by an assertion
34879 check. Add assertions to prevent ICE if attempts are made to
34880 define an altivec predicate or an abs* built-in function that has
34882 (htm_init_builtins): Add comment explaining why definition of the
34883 htm built-in functions is not preceded by an assertion check.
34885 2017-01-04 Jeff Law <law@redhat.com>
34887 PR tree-optimizatin/67955
34888 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34889 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34890 the points-to solution does not include pt_null. Use DECL_PT_UID
34893 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34895 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34896 Use gen_int_mode instead of gen_lopwart for const_int operands.
34898 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34900 PR tree-optimization/71563
34901 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34902 out of range value - has low bits known to be zero.
34904 2017-01-04 Alan Modra <amodra@gmail.com>
34906 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34907 * configure: Regenerate.
34908 * config.in: Regenerate.
34910 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34913 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34914 a substring of the message, but strcmp with the whole message. Ifdef
34915 ENABLE_NLS, translate the message first using dgettext.
34917 2017-01-03 Jeff Law <law@redhat.com>
34919 PR tree-optimizatin/78856
34920 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34921 (mark_threaded_blocks): Remove code to truncate thread paths that
34922 cross multiple loop headers. Instead invalidate the cached loop
34923 iteration information and handle case of a thread path walking
34924 into an irreducible region.
34926 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34929 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34930 assertions. Add support for doing the signbit if the IEEE 128-bit
34931 floating point value is in a GPR.
34932 * config/rs6000/rs6000.md (Fsignbit): Delete.
34933 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34934 Update the length attribute if the value is in a GPR.
34935 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34936 the sign or zero extension instruction, since the value is always 0/1.
34937 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34940 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34941 extracting SImode to a GPR register so that we can generate a
34942 store, limit the vector to be in a traditional Altivec register
34943 for the vextuwrx instruction.
34945 2017-01-03 Ian Lance Taylor <iant@google.com>
34947 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34949 2017-01-03 Martin Sebor <msebor@redhat.com>
34951 PR tree-optimization/78696
34952 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34953 precision. Use MPFR for %f for greater fidelity. Correct handling
34955 (pass_sprintf_length::compute_format_length): Set width and precision
34956 specified by asrerisk to void_node for vararg functions.
34957 (try_substitute_return_value): Adjust dump output.
34959 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34961 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34963 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34965 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34966 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34967 -mlra/-mno-lra was passed to the compiler.
34969 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34971 PR rtl-optimization/65618
34972 * emit-rtl.c (try_split): Move initialization of "before" and
34973 "after" to just before the call to emit_insn_after_setloc.
34975 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34977 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34979 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34981 * dwarf2out.c (gen_enumeration_type_die): When
34982 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34984 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34986 PR tree-optimization/78965
34987 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34988 Change first argument from const call_info & to call_info &. For %n
34989 set info.nowrite to false.
34991 PR middle-end/78901
34992 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34993 possibly throwing calls.
34995 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34996 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34997 and fns handling, rather than in a separate case SSA_NAME.
34999 2017-01-02 Jeff Law <law@redhat.com>
35001 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35002 fixes for first_period and second_period variables.
35004 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35007 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35008 (*insvqi_1): New insn pattern.
35009 (*insvqi_1_mem_rex64): Ditto.
35010 (*insvqi_2): Ditto.
35011 (*insvqi_3): Rename from *insvqi.
35013 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35015 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35017 * doc/cfg.texi (Edges): Remove reference to Java.
35018 (Maintaining the CFG): Ditto.
35020 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35022 PR middle-end/77674
35023 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35024 transparent aliases.
35026 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35028 PR middle-end/77484
35029 * predict.def (PRED_CALL): Update hitrate.
35030 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35031 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35032 into direct/indirect/polymorphic variants.
35034 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35036 Update copyright years.
35038 * gcc.c (process_command): Update copyright notice dates.
35039 * gcov-dump.c (print_version): Ditto.
35040 * gcov.c (print_version): Ditto.
35041 * gcov-tool.c (print_version): Ditto.
35042 * gengtype.c (create_file): Ditto.
35043 * doc/cpp.texi: Bump @copying's copyright year.
35044 * doc/cppinternals.texi: Ditto.
35045 * doc/gcc.texi: Ditto.
35046 * doc/gccint.texi: Ditto.
35047 * doc/gcov.texi: Ditto.
35048 * doc/install.texi: Ditto.
35049 * doc/invoke.texi: Ditto.
35051 Copyright (C) 2017 Free Software Foundation, Inc.
35053 Copying and distribution of this file, with or without modification,
35054 are permitted in any medium without royalty provided the copyright
35055 notice and this notice are preserved.