1 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3 * common.opt (gcoff): Re-add as ignored option.
4 (gcoff1, gcoff2, gcoff3): Likewise.
6 * Makefile.in (OBJS): Delete sdbout.o.
7 (GTFILES): Delete $(srcdir)/sdbout.c.
8 * debug.h: Delete sdb_debug_hooks.
9 * final.c: Delete sdbout.h include.
10 (final_scan_insn): Delete SDB_DEBUG check.
11 (rest_of_clean_state): Likewise.
12 * output.h: Delete sdb_begin_function_line.
15 * toplev.c: Delete sdbout.h include.
16 (process_options): Delete SDB_DEBUG check.
17 * tree-core.h (tree_type_common): Delete pointer field of
19 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
21 * tree.h (TYPE_SYMTAB_POINTER): Delete.
22 (TYPE_SYMTAB_IS_POINTER): Delete.
23 (TYPE_SYMTAB_IS_DIE): Renumber.
24 * xcoffout.c: Refer to former sdbout.c file.
25 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
27 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
28 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
29 (Debugging Options): Delete -gcoff.
30 (-gstabs): Delete SDB reference.
32 (-gcoff@var{level}): Delete.
33 * doc/passes.texi (Debugging information output): Delete SDB and
35 * doc/tm.texi: Regenerate.
36 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
37 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
39 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
40 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
41 (SDB_DEBUGGING_INFO): Delete.
42 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
43 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
44 * target.def (output_source_filename): Delete COFF reference.
46 * common.opt (gcoff): Delete.
47 (gxcoff+): Update Negative chain.
48 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
50 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
51 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
52 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
54 (expand_function_start): Change sdb reference to past tense.
55 (expand_function_end): Change sdb reference to past tense.
56 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
57 * opts.c (debug_type_names): Delete coff entry.
58 (common_handle_option): Delete OPT_gcoff case.
59 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
61 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
62 * config/cris/cris.h: Delete SDB reference in comment.
63 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
64 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
65 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
66 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
68 (ix86_expand_prologue): Likewise.
69 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
70 * config/ia64/ia64.h: Likewise.
71 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
72 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
74 * config/mmix/mmix.h: Likewise.
75 * config/nds32/nds32.c: Likewise.
76 * config/stormy/storym16.h: Likewise.
77 * config/visium/visium.h: Likewise.
78 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
80 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
82 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
83 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
84 FRAME_POINTER_REGNUM point at high end of local var area.
86 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
88 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
89 Move comment around. Do not reset best_edge for a copiable
90 destination if the copy would cause a partition change.
91 (better_edge_p): Remove redundant check.
93 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
95 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
97 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
99 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
100 for math functions that have _Float<N> and _Float<N>X variants.
101 (mathfn_built_in_2): Add support for math functions that have
102 _Float<N> and _Float<N>X variants.
103 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
104 (expand_builtin_mathfn_ternary): Add support for fma with
105 _Float<N> and _Float<N>X variants.
106 (expand_builtin): Likewise.
107 (fold_builtin_3): Likewise.
108 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
109 create math function _Float<N> and _Float<N>X variants as external
111 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
112 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
113 the __builtin_ prefix and if not strict ansi, without the prefix.
114 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
115 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
116 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
117 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
118 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
119 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
120 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
121 function signatures for fma _Float<N> and _Float<N>X variants.
122 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
123 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
124 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
125 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
126 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
127 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
128 * gencfn-macros.c (print_case_cfn): Add support for math functions
129 that have _Float<N> and _Float<N>X variants.
130 (print_define_operator_list): Likewise.
131 (fltfn_suffixes): Likewise.
133 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
134 for math functions that have _Float<N> and _Float<N>X variants.
135 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
136 and _Float<N>X variants.
137 (COPYSIGN): Likewise.
140 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
141 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
143 (integer_valued_read_call_p): Likewise.
144 * fold-const-call.c (fold_const_call_ss): Likewise.
145 (fold_const_call_sss): Add support for copysign, fmin, and fmax
146 _Float<N> and _Float<N>X variants.
147 (fold_const_call_ssss): Add support for fma _Float<N> and
149 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
150 support for copysign and fma _Float<N> and _Float<N>X variants.
151 (backprop::process_builtin_call_use): Likewise.
152 * tree-call-cdce.c (can_test_argument_range); Add support for
153 sqrt _Float<N> and _Float<N>X variants.
154 (edom_only_function): Likewise.
155 (get_no_error_domain): Likewise.
156 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
157 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
158 copysign _Float<N> and _Float<N>X variants.
159 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
160 handled by machine independent code.
162 * doc/cpp.texi (Common Predefined Macros): Document defining
163 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
164 fma _Float<N> and _Float<N>X variants.
166 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
169 * config/i386/i386-modes.def (CCFPU): Remove definition.
170 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
171 (ix86_cc_modes_compatible): Ditto.
172 (ix86_expand_carry_flag_compare): Ditto.
173 (ix86_expand_int_movcc): Ditto.
174 (ix86_expand_int_addcc): Ditto.
175 (ix86_reverse_condition): Ditto.
176 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
177 Return true/false for unordered/ordered fp comparisons.
178 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
179 (ix86_prepare_fp_compare_args): Update for rename.
180 (ix86_expand_fp_compare): Update for rename. Generate unordered
181 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
182 (ix86_expand_sse_compare_and_jump): Ditto.
183 * config/i386/predicates.md (fcmov_comparison_operator):
184 Remove CCFPU mode handling.
185 (ix86_comparison_operator): Ditto.
186 (ix86_carry_flag_operator): Ditto.
187 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
188 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
189 (*cmpu<mode>_cc_i387): Ditto.
190 (FPCMP): Remove mode iterator.
191 (unord): Remove mode attribute.
192 (unord_subst): New define_subst transformation
193 (unord): New define_subst attribute.
195 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
196 (*cmpi<unord>xf_i387): Ditto.
197 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
198 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
199 using unord_subst transformation.
200 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
201 (round_saeonly): Also handle CCFP mode.
202 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
203 Remove UNSPEC_SAHF unspec handling.
205 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
207 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
209 2017-10-27 Jeff Law <law@redhat.com>
211 * gimple-ssa-sprintf.c: Include domwalk.h.
212 (class sprintf_dom_walker): New class, derived from dom_walker.
213 (sprintf_dom_walker::before_dom_children): New function.
214 (struct call_info): Moved into sprintf_dom_walker class
215 (compute_formath_length, handle_gimple_call): Likewise.
216 (sprintf_length::execute): Call the dominator walker rather
217 than walking the statements.
219 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
220 gimple statement locations.
221 (check_array_bounds): Corresponding changes. Get the statement's
222 location directly from wi->stmt.
224 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
227 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
229 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
231 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
232 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
234 2017-10-27 Jakub Jelinek <jakub@redhat.com>
237 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
238 * config/i386/i386.c (maybe_get_pool_constant): Removed.
239 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
240 maybe_get_pool_constant.
241 * config/i386/predicates.md (zero_extended_scalar_load_operand):
244 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
246 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
247 2.26 caveat. Update gas and gld versions.
248 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
251 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
253 * cgraph.h (set_malloc_flag): Declare.
254 * cgraph.c (set_malloc_flag_1): New function.
255 (set_malloc_flag): Likewise.
256 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
257 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
259 (read_ipa_call_summary): Add support for reading is_return_callee.
260 (write_ipa_call_summary): Stream is_return_callee.
261 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
262 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
263 ipa-prop.h, ipa-fnsummary.h.
264 (pure_const_names): Change to static.
265 (malloc_state_e): Define.
266 (malloc_state_names): Define.
267 (funct_state_d): Add field malloc_state.
268 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
269 (check_retval_uses): New function.
270 (malloc_candidate_p): Likewise.
271 (analyze_function): Add support for malloc attribute.
272 (pure_const_write_summary): Stream malloc_state.
273 (pure_const_read_summary): Add support for reading malloc_state.
274 (dump_malloc_lattice): New function.
275 (propagate_malloc): New function.
276 (warn_function_malloc): New function.
277 (ipa_pure_const::execute): Call propagate_malloc and
279 (pass_local_pure_const::execute): Add support for malloc attribute.
280 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
281 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
283 2017-10-27 Martin Liska <mliska@suse.cz>
285 PR gcov-profile/82457
286 * doc/invoke.texi: Document that one needs a non-strict ISO mode
287 for fork-like functions to be properly instrumented.
289 2017-10-27 Richard Biener <rguenther@suse.de>
292 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
293 info when we redirected EH.
295 2017-10-26 Michael Collison <michael.collison@arm.com>
297 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
299 (<optab>_trunchf<GPI:mode>2: New pattern.
300 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
301 * config/aarch64/iterators.md (wv): New mode attribute.
302 (vf, VF): New mode attributes.
303 (vgp, VGP): New mode attributes.
304 (s): Update attribute with SImode and DImode prefixes.
306 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
308 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
309 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
310 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
311 (nios2_option_overide): Initialize it. Don't allow R0-relative
313 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
314 (nios2_symbolic_constant_p): Likewise.
315 (nios2_legitimate_address_p): Likewise.
316 (nios2_r0rel_section_name_p): New.
317 (nios2_symbol_ref_in_r0rel_data_p): New.
318 (nios2_emit_move_sequence): Handle r0rel_constant_p.
319 (r0rel_constant_p): New.
320 (nios2_print_operand_address): Handle r0rel_constant_p.
321 (nios2_cdx_narrow_form_p): Likewise.
322 * config/nios2/nios2.opt (mr0rel-sec=): New option.
323 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
324 (Nios II Options): Document -mr0rel-sec.
326 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
328 * config/nios2/nios2.c: Include xregex.h.
329 (nios2_gprel_sec_regex): New.
330 (nios2_option_overide): Initialize it. Don't allow GP-relative
332 (nios2_small_section_name_p): Check for regex match.
333 * config/nios2/nios2.opt (mgprel-sec=): New option.
334 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
335 (Nios II Options): Document -mgprel-sec.
337 2017-10-26 Jim Wilson <wilson@tuliptree.org>
339 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
341 2017-10-26 Tom de Vries <tom@codesourcery.com>
343 PR tree-optimization/82707
344 * gimple.c (gimple_copy): Fix unsharing of
345 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
347 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
349 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
350 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
351 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
352 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
353 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
354 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
355 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
356 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
357 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
358 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
359 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
360 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
361 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
362 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
363 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
364 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
365 _mm512_mask_cmpunord_ps_mask): New intrinsics.
367 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
369 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
371 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
372 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
373 warning to rs6000.c. Remove the Undocumented flag, since it has
375 (-mabi=ibmlongdouble): Likewise.
376 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
377 already set, set the default format for long double.
378 (rs6000_debug_reg_global): Print whether long double is IBM or
380 (rs6000_option_override_internal): Rework setting long double
381 format. Only warn if the user is changing the long double default
382 and they did not use -Wno-psabi.
383 * doc/invoke.texi (PowerPC options): Update the documentation for
384 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
386 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
387 Alan Hayward <alan.hayward@arm.com>
388 David Sherwood <david.sherwood@arm.com>
390 * rtl.h (wider_subreg_mode): New function.
391 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
392 rather than an unsigned int *.
393 * ira-color.c (regno_max_ref_width): Replace with...
394 (regno_max_ref_mode): ...this new variable.
395 (coalesced_pseudo_reg_slot_compare): Update accordingly.
396 Use wider_subreg_mode.
397 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
398 rather than an unsigned int *.
399 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
400 (process_alt_operands): Likewise.
401 (invariant_p): Likewise.
402 * lra-spills.c (assign_mem_slot): Likewise.
403 (add_pseudo_to_slot): Likewise.
404 * lra.c (collect_non_operand_hard_regs): Likewise.
405 (add_regs_to_insn_regno_info): Likewise.
406 * reload1.c (regno_max_ref_width): Replace with...
407 (regno_max_ref_mode): ...this new variable.
408 (reload): Update accordingly. Update call to
409 ira_sort_regnos_for_alter_reg.
410 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
411 (init_eliminable_invariants): Update to use regno_max_ref_mode.
412 (scan_paradoxical_subregs): Likewise.
414 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
416 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
417 (aarch64_frame): Add emit_frame_chain boolean.
418 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
419 Move eh_return case to aarch64_layout_frame.
420 (aarch64_layout_frame): Initialize emit_frame_chain.
421 (aarch64_expand_prologue): Use emit_frame_chain.
423 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
425 * config/aarch64/aarch64.c (aarch64_layout_frame):
426 Ensure LR is always stored at the bottom of the callee-saves.
427 Remove rarely used frame layout which saves callee-saves at top of
428 frame, so the store of LR can be used as a valid probe in all cases.
430 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
432 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
433 Improve unaligned TImode/TFmode base/offset split.
435 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
436 Alan Hayward <alan.hayward@arm.com>
437 David Sherwood <david.sherwood@arm.com>
439 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
440 * combine.c (find_single_use_1): Likewise.
441 (expand_field_assignment): Likewise.
442 (move_deaths): Likewise.
443 * lra-constraints.c (simplify_operand_subreg): Likewise.
444 (curr_insn_transform): Likewise.
445 * lra.c (collect_non_operand_hard_regs): Likewise.
446 (add_regs_to_insn_regno_info): Likewise.
447 * rtlanal.c (reg_referenced_p): Likewise.
448 (covers_regno_no_parallel_p): Likewise.
450 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
452 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
453 Don't print any bits outside the precision of the value.
454 * wide-int.cc (test_printing): Add some new tests.
456 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
458 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
459 supports -xbrace_comment option.
460 * configure: Regenerate.
461 * config.in: Regenerate.
462 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
463 (ASM_CPU_SPEC): Use it.
465 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
467 * target.def (static_rtx_alignment): New hook.
468 * targhooks.h (default_static_rtx_alignment): Declare.
469 * targhooks.c (default_static_rtx_alignment): New function.
470 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
471 * doc/tm.texi: Regenerate.
472 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
473 instead of targetm.constant_alignment. Remove call to
475 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
476 (cris_preferred_mininum_alignment): New function, split out from...
477 (cris_constant_alignment): ...here.
478 (cris_static_rtx_alignment): New function.
479 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
481 (ix86_constant_alignment): ...here.
482 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
483 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
484 (mmix_static_rtx_alignment): New function.
485 * config/spu/spu.c (spu_static_rtx_alignment): New function.
486 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
488 2017-10-26 Tamar Christina <tamar.christina@arm.com>
491 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
492 Add flag_trapping_math and flag_fp_int_builtin_inexact.
494 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
496 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
497 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
498 mark as a sign-extending load.
499 (local_pic_load_u): Define.
501 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
504 * fold-const.c (operand_equal_for_comparison_p): Also return true
505 if ARG0 is a simple variant of ARG1 with narrower precision.
506 (fold_ternary_loc): Always pass unstripped operands to the predicate.
508 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
510 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
512 * i386.h (processor_costs): Add gather_static, gather_per_elt,
513 scatter_static, scatter_per_elt.
514 * x86-tune-costs.h: Add new cost entries.
516 2017-10-25 Richard Biener <rguenther@suse.de>
518 * tree-ssa-sccvn.h (vn_eliminate): Declare.
519 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
520 class pass_fre): Move to ...
521 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
522 class pass_fre): ... here and adjust for statistics.
524 2017-10-25 Jakub Jelinek <jakub@redhat.com>
527 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
528 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
529 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
530 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
533 2017-10-25 Richard Biener <rguenther@suse.de>
535 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
536 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
537 eliminate_push_avail, eliminate_insert): Move inside...
538 (class eliminate_dom_walker): ... this class in preparation
540 (fini_eliminate): Remove by merging with ...
541 (eliminate): ... this function. Adjust for class changes.
542 (pass_pre::execute): Remove fini_eliminate call.
543 (pass_fre::execute): Likewise.
545 2017-10-24 Jakub Jelinek <jakub@redhat.com>
548 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
549 (VPERMI2, VPERMI2I): New mode iterators.
550 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
551 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
553 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
554 mode iterator. Remove 3 old define_insn patterns.
555 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
556 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
557 VPERMI2 mode iterator, remove the other two expanders.
558 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
559 to use VPERMI2 mode iterator, add another alternative for vpermi2*
560 instructions, remove the other two patterns.
561 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
562 mode iterator, remove the other two patterns.
563 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
564 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
565 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
566 and adjust argument order accordingly.
567 (ix86_expand_vec_perm): Adjust caller.
568 (expand_vec_perm_1): Likewise.
569 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
570 (expand_vec_perm_vpermt2_vpshub2): ... this.
571 (ix86_expand_vec_perm_const_1): Adjust caller.
572 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
575 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
576 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
577 (vec_shl_<mode>): Remove unused expander.
578 (avx512bw_<shift_insn><mode>3): New define_insn.
579 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
580 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
582 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
585 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
588 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
590 PR rtl-optimization/82396
591 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
592 (autopref_multipass_init): Simplify initialization.
593 (autopref_rank_data): Simplify sort order.
594 * gcc/sched-int.h (autopref_multipass_data_): Remove
595 multi_mem_insn_p, min_offset and max_offset.
597 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
600 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
601 Check special value of flag_omit_frame_pointer.
602 (aarch64_can_eliminate): Likewise.
603 (aarch64_override_options_after_change_1): Simplify handling of
604 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
606 2017-10-24 Richard Biener <rguenther@suse.de>
608 PR tree-optimization/82697
609 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
610 zero for conditional load and unconditional store.
612 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
614 * doc/install.texi: Document bootstrap-cet.
616 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
619 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
620 ENDBR instruction at function entrance if function is only
623 2017-10-24 Jakub Jelinek <jakub@redhat.com>
626 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
627 patterns to better describe from which operation the CF is computed.
628 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
629 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
630 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
631 is 0, use _0 suffixed expanders instead of emitting a comparison
634 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
636 * config/i386/i386.md(*movsf_internal, *movdf_internal):
637 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
639 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
642 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
643 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
644 * loop-iv.c (iv_get_reaching_def): Likewise.
645 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
646 variable is promoted and the partition contains undefined values.
648 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
650 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
652 (nios2_address_cost): Define.
653 (nios2_legitimize_address): Recognize (exp + constant) directly.
654 (TARGET_ADDRESS_COST): Define.
656 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
658 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
659 (nios2_symbolic_memory_operand_p): Declare.
660 (nios2_split_large_constant): Declare.
661 (nios2_split_symbolic_memory_operand): Declare.
662 * config/nios2/nios2.c: Adjust includes.
663 (nios2_symbolic_constant_allowed): New.
664 (nios2_symbolic_constant_p): New.
665 (nios2_plus_symbolic_constant_p): New.
666 (nios2_valid_addr_expr_p): Recognize addresses involving
668 (nios2_legitimate_address_p): Likewise, also LO_SUM.
669 (nios2_symbolic_memory_operand_p): New.
670 (nios2_large_constant_p): New.
671 (nios2_split_large_constant): New.
672 (nios2_split_plus_large_constant): New.
673 (nios2_split_symbolic_memory_operand): New.
674 (nios2_legitimize_address): Code refactoring. Handle addresses
675 involving symbolic constants.
676 (nios2_emit_move_sequence): Likewise.
677 (nios2_print_operand): Improve error output.
678 (nios2_print_operand_address): Handle LO_SUM.
679 (nios2_cdx_narrow_form_p): Likewise.
680 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
681 operands involving symbolic constants.
682 (movhi_internal, movsi_internal): Likewise.
683 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
684 (extendhisi2, extendqi<mode>2): Likewise.
686 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
688 * tree-pass.h (PROP_rtl_split_insns): Define.
689 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
691 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
693 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
695 2017-10-23 Jakub Jelinek <jakub@redhat.com>
698 * target.def (const_not_ok_for_debug_p): Default to
699 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
700 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
701 * targhooks.c (default_const_not_ok_for_debug_p): New function.
702 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
703 which targetm.const_not_ok_for_debug_p returned true.
704 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
706 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
708 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
709 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
710 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
712 (ix86_const_not_ok_for_debug_p): New function.
713 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
714 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
716 2017-10-23 David Malcolm <dmalcolm@redhat.com>
719 * system.h: Conditionally include "unique-ptr.h" if
720 INCLUDE_UNIQUE_PTR is defined.
721 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
722 of defining INCLUDE_UNIQUE_PTR before including "system.h".
724 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
726 * config/rl78/rl78.md: New define_expand "subdi3".
728 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
731 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
732 DF_REF_INSN if DF_REF_INSN_INFO is false.
734 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
736 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
737 xmm_move instead of sse_move.
738 (sse_store_index): New function.
739 (ix86_register_move_cost): Be more sensible about mismatch stall;
740 model AVX moves correctly; make difference between sse->integer and
742 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
743 moves; make difference between SSE and AVX.
744 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
745 and zmm_move. Increase size of sse load and store tables;
746 add unaligned load and store tables; add ssemmx_to_integer.
747 * x86-tune-costs.h: Update all entries according to real
748 move latencies from Agner Fog's manual and chip documentation.
750 2017-10-23 Jakub Jelinek <jakub@redhat.com>
753 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
754 * config/i386/constraints.md (Wf): New constraint.
755 * config/i386/i386.md (UNSPEC_SBB): New unspec.
756 (cmp<dwi>_doubleword): Removed.
757 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
758 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
759 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
760 expand with cmp<dwi>_doubleword. For LTU and GEU use
761 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
763 * common.opt (gcolumn-info): Enable by default.
764 * doc/invoke.texi (gcolumn-info): Document new default.
766 2017-10-23 Richard Biener <rguenther@suse.de>
768 PR tree-optimization/82672
769 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
770 Fold the stmt if we propagated into it.
772 2017-10-23 Richard Biener <rguenther@suse.de>
774 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
775 (bitmap_remove_expr_from_set): ... this. All callers call this
776 for non-constant values.
777 (bitmap_set_subtract): Rename to...
778 (bitmap_set_subtract_expressions): ... this. Adjust and
780 (bitmap_set_contains_value): Remove superfluous check.
781 (bitmap_set_replace_value): Inline into single caller ...
782 (bitmap_value_replace_in_set): ... here and simplify.
783 (dependent_clean): Merge into ...
784 (clean): ... this using an overload. Adjust.
785 (prune_clobbered_mems): Adjust.
786 (compute_antic_aux): Likewise.
787 (compute_partial_antic_aux): Likewise.
789 2017-10-23 Richard Biener <rguenther@suse.de>
791 PR tree-optimization/82129
793 2017-08-01 Richard Biener <rguenther@suse.de>
795 PR tree-optimization/81181
796 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
797 (compute_antic): ... end of iteration here.
799 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
801 * target.def (starting_frame_offset): New hook.
802 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
803 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
804 * doc/tm.texi.in: Regenerate.
805 * hooks.h (hook_hwi_void_0): Declare.
806 * hooks.c (hook_hwi_void_0): New function.
807 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
808 STARTING_FRAME_OFFSET.
809 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
810 * reload1.c (reload): Likewise.
811 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
812 instead of STARTING_FRAME_OFFSET.
813 * function.c (try_fit_stack_local): Likewise.
814 (assign_stack_local_1): Likewise
815 (instantiate_virtual_regs): Likewise.
816 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
817 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
818 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
819 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
820 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
821 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
822 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
823 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
824 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
825 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
826 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
827 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
828 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
829 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
830 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
831 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
832 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
833 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
834 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
835 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
836 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
837 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
838 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
839 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
840 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
841 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
842 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
843 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
844 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
845 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
846 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
847 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
848 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
849 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
850 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
851 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
852 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
853 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
854 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
855 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
856 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
857 * config/avr/avr.c (avr_starting_frame_offset): Make static and
858 return a HOST_WIDE_INT.
859 (avr_builtin_setjmp_frame_value): Use it instead of
860 STARTING_FRAME_OFFSET.
861 (TARGET_STARTING_FRAME_OFFSET): Redefine.
862 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
863 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
865 (TARGET_STARTING_FRAME_OFFSET): Redefine.
866 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
867 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
868 (TARGET_CONSTANT_ALIGNMENT): Redefine.
869 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
870 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
871 (TARGET_STARTING_FRAME_OFFSET): Redefine.
872 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
873 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
874 (TARGET_STARTING_FRAME_OFFSET): Redefine.
875 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
876 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
878 (TARGET_STARTING_FRAME_OFFSET): Redefine.
879 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
880 * config/mips/mips.c (mips_compute_frame_info): Refer to
881 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
882 (mips_starting_frame_offset): New function.
883 (TARGET_STARTING_FRAME_OFFSET): Redefine.
884 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
885 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
886 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
887 and return a HOST_WIDE_INT.
888 (TARGET_STARTING_FRAME_OFFSET): Redefine.
889 (mmix_initial_elimination_offset): Refer to
890 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
891 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
892 * config/pa/pa.c (pa_starting_frame_offset): New function.
893 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
894 (pa_expand_prologue): Likewise.
895 (TARGET_STARTING_FRAME_OFFSET): Redefine.
896 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
897 !FRAME_GROWS_DOWNWARD handling to...
898 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
899 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
900 !FRAME_GROWS_DOWNWARD handling to...
901 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
902 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
903 !FRAME_GROWS_DOWNWARD handling to...
904 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
905 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
907 (rs6000_starting_frame_offset): New function.
908 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
909 !FRAME_GROWS_DOWNWARD handling to...
910 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
911 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
912 !FRAME_GROWS_DOWNWARD handling to...
913 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
914 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
915 !FRAME_GROWS_DOWNWARD handling to...
916 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
917 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
918 (rs6000_starting_frame_offset): New function.
919 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
920 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
921 * config/vax/vax.c (vax_starting_frame_offset): New function.
922 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
923 (TARGET_STARTING_FRAME_OFFSET): Redefine.
924 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
925 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
926 (TARGET_STARTING_FRAME_OFFSET): Redefine.
927 * system.h (STARTING_FRAME_OFFSET): Poison.
929 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
931 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
932 SCALAR_TYPE_MODE instead of TYPE_MODE.
934 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
935 Alan Hayward <alan.hayward@arm.com>
936 David Sherwood <david.sherwood@arm.com>
938 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
940 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
941 Alan Hayward <alan.hayward@arm.com>
942 David Sherwood <david.sherwood@arm.com>
944 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
946 2017-10-23 Richard Biener <rguenther@suse.de>
948 PR tree-optimization/82129
949 * tree-ssa-pre.c (bitmap_set_and): Remove.
950 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
951 canonicalizing expressions in the set to those with lowest
952 ID rather than taking that from the first edge.
954 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
956 * combine.c (rtx_equal_for_field_assignment_p): Use
959 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
960 Alan Hayward <alan.hayward@arm.com>
961 David Sherwood <david.sherwood@arm.com>
963 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
964 to a SUBREG_PROMOTED_VAR.
966 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
967 Alan Hayward <alan.hayward@arm.com>
968 David Sherwood <david.sherwood@arm.com>
970 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
971 (expand_debug_source_expr): Likewise.
972 * combine.c (combine_simplify_rtx): Likewise.
973 * cse.c (fold_rtx): Likewise.
974 * optabs.c (expand_float): Likewise.
975 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
976 (simplify_binary_operation_1): Likewise.
978 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
979 Alan Hayward <alan.hayward@arm.com>
980 David Sherwood <david.sherwood@arm.com>
982 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
983 (record_promoted_value): Likewise.
984 * expr.c (expand_expr_real_2): Likewise.
985 * ree.c (update_reg_equal_equiv_notes): Likewise.
986 (combine_set_extension): Likewise.
987 * rtlanal.c (low_bitmask_len): Likewise.
988 * simplify-rtx.c (neg_const_int): Likewise.
989 (simplify_binary_operation_1): Likewise.
991 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
992 Alan Hayward <alan.hayward@arm.com>
993 David Sherwood <david.sherwood@arm.com>
995 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
996 * regcprop.c (maybe_mode_change): Likewise.
997 * reload1.c (alter_reg): Likewise.
999 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1001 * inchash.h (inchash::hash::add_wide_int): New function.
1002 * lto-streamer-out.c (hash_tree): Use it.
1004 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1006 * inchash.h (inchash::hash::add_wide_int): Rename to...
1007 (inchash::hash::add_hwi): ...this.
1008 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
1009 (polymorphic_call_target_hasher::hash): Likewise.
1010 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
1011 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
1012 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
1013 * lto-streamer-out.c (hash_tree): Likewise.
1014 * optc-save-gen.awk: Likewise.
1015 * tree.c (add_expr): Likewise.
1017 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1020 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
1021 for ordered inequality comparisons even with TARGET_IEEE_FP.
1023 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1026 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
1027 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1028 Expand with cmp<dwi>_doubleword.
1030 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1032 * extend.texi: Add x86 specific to 'nocf_check' attribute.
1033 List CET intrinsics.
1034 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
1035 specific to -fcf-protection option.
1037 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1039 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
1040 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
1041 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
1042 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
1043 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
1044 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
1045 (extra_objs): Add cet.o for Linux/x86 targets.
1046 (tmake_file): Add i386/t-cet for Linux/x86 targets.
1047 * config/i386/cet.c: New file.
1048 * config/i386/cetintrin.h: Likewise.
1049 * config/i386/t-cet: Likewise.
1050 * config/i386/cpuid.h (bit_SHSTK): New.
1051 (bit_IBT): Likewise.
1052 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
1053 pass IBT and SHSTK bits.
1054 * config/i386/i386-builtin-types.def
1055 (VOID_FTYPE_UNSIGNED_PVOID): New.
1056 (VOID_FTYPE_UINT64_PVOID): Likewise.
1057 * config/i386/i386-builtin.def: Add CET intrinsics.
1058 * config/i386/i386-c.c (ix86_target_macros_internal): Add
1059 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
1060 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
1061 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
1063 * config/i386/i386.c (rest_of_insert_endbranch): New.
1064 (pass_data_insert_endbranch): Likewise.
1065 (pass_insert_endbranch): Likewise.
1066 (make_pass_insert_endbranch): Likewise.
1067 (ix86_notrack_prefixed_insn_p): Likewise.
1068 (ix86_target_string): Add -mibt, -mshstk flags.
1069 (ix86_option_override_internal): Add flag_cf_protection
1071 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
1072 (ix86_print_operand): Add 'notrack' prefix output.
1073 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
1074 (ix86_expand_builtin): Expand CET intrinsics.
1075 (x86_output_mi_thunk): Add 'endbranch' instruction.
1076 * config/i386/i386.h (TARGET_IBT): New.
1077 (TARGET_IBT_P): Likewise.
1078 (TARGET_SHSTK): Likewise.
1079 (TARGET_SHSTK_P): Likewise.
1080 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
1081 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
1082 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
1083 (builtin_setjmp_setup): New pattern.
1084 (builtin_longjmp): Likewise.
1085 (rdssp<mode>): Likewise.
1086 (incssp<mode>): Likewise.
1087 (saveprevssp): Likewise.
1088 (rstorssp): Likewise.
1089 (wrss<mode>): Likewise.
1090 (wruss<mode>): Likewise.
1091 (setssbsy): Likewise.
1092 (clrssbsy): Likewise.
1093 (nop_endbr): Likewise.
1094 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
1096 * config/i386/immintrin.h: Include <cetintrin.h>.
1097 * config/i386/linux-common.h
1098 (file_end_indicate_exec_stack_and_cet): New prototype.
1099 (TARGET_ASM_FILE_END): New.
1101 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
1103 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
1104 latencies instead of having separate table; make difference between
1105 integer and float costs.
1106 * i386.h (processor_costs): Remove scalar_stmt_cost,
1107 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
1108 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
1110 * x86-tune-costs.h: Remove entries which has been removed in
1111 procesor_costs from all tables; make cond_taken_branch_cost
1112 and cond_not_taken_branch_cost COST_N_INSNS based.
1114 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
1116 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
1118 2017-10-20 Jakub Jelinek <jakub@redhat.com>
1120 * config/i386/i386.md (isa): Remove fma_avx512f.
1121 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
1122 <avx512>_fmadd_<mode>_mask3<round_name>,
1123 <avx512>_fmsub_<mode>_mask<round_name>,
1124 <avx512>_fmsub_<mode>_mask3<round_name>,
1125 <avx512>_fnmadd_<mode>_mask<round_name>,
1126 <avx512>_fnmadd_<mode>_mask3<round_name>,
1127 <avx512>_fnmsub_<mode>_mask<round_name>,
1128 <avx512>_fnmsub_<mode>_mask3<round_name>,
1129 <avx512>_fmaddsub_<mode>_mask<round_name>,
1130 <avx512>_fmaddsub_<mode>_mask3<round_name>,
1131 <avx512>_fmsubadd_<mode>_mask<round_name>,
1132 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
1133 (*vec_widen_umult_even_v16si<mask_name>,
1134 *vec_widen_smult_even_v16si<mask_name>): Likewise.
1135 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
1137 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1139 * extend.texi: Add 'nocf_check' documentation.
1140 * gimple.texi: Add second parameter to
1141 gimple_build_call_from_tree.
1142 * invoke.texi: Add -fcf-protection documentation.
1143 * rtl.texi: Add REG_CALL_NOTRACK documenation.
1145 2017-10-20 Richard Biener <rguenther@suse.de>
1147 PR tree-optimization/82473
1148 * tree-vect-loop.c (vectorizable_reduction): Properly get at
1149 the largest input type.
1151 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1153 * c-attribs.c (handle_nocf_check_attribute): New function.
1154 (c_common_attribute_table): Add 'nocf_check' handling.
1155 * gimple-parser.c: Add second argument NULL to
1156 gimple_build_call_from_tree.
1157 * attrib.c (comp_type_attributes): Check nocf_check attribute.
1158 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
1160 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
1161 * common.opt: Add fcf-protection flag.
1162 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
1163 * flag-types.h: Add enum cf_protection_level.
1164 * gimple.c (gimple_build_call_from_tree): Add second parameter.
1165 Add 'nocf_check' attribute propagation to gimple call.
1166 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
1167 (gimple_build_call_from_tree): Update prototype.
1168 (gimple_call_nocf_check_p): New function.
1169 (gimple_call_set_nocf_check): Likewise.
1170 * gimplify.c: Add second argument to gimple_build_call_from_tree.
1171 * ipa-icf.c: Add nocf_check attribute in statement hash.
1172 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
1173 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
1174 * toplev.c (process_options): Add flag_cf_protection handling.
1176 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1178 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
1180 2017-10-20 Richard Biener <rguenther@suse.de>
1182 PR tree-optimization/82603
1183 * tree-if-conv.c (predicate_mem_writes): Make sure to only
1184 remove false predicated stores.
1186 2017-10-20 Richard Biener <rguenther@suse.de>
1188 * graphite-isl-ast-to-gimple.c
1189 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
1190 Remove return value and simplify, dump copied stmt after lhs
1192 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
1193 Reduce dump verbosity.
1194 (gsi_insert_earliest): Likewise.
1195 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
1196 * graphite.c (print_global_statistics): Adjust dumping.
1197 (print_graphite_scop_statistics): Likewise.
1198 (print_graphite_statistics): Do not dump loops here.
1199 (graphite_transform_loops): But here.
1201 2017-10-20 Nicolas Roche <roche@adacore.com>
1203 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
1204 * configure: Regenerate.
1206 2017-10-20 Jakub Jelinek <jakub@redhat.com>
1209 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
1210 functions when optimizing replace GIMPLE_RETURN stmts with
1211 calls to __builtin_unreachable ().
1214 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
1215 for -fsanitize=thread link of executables.
1216 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
1217 link of executables.
1220 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
1222 (<shift_insn><mode>3<mask_name>): Change the last of the 3
1223 define_insns for logical vector shifts to use VI248_AVX512BW
1224 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
1225 condition, useless isa and prefix attributes. Change the first
1226 2 of these define_insns to ...
1227 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
1228 define_insn for avx512vl.
1229 (<shift_insn><mode>3): ... and this, new define_insn without
1230 masking for non-avx512vl.
1233 * config/i386/sse.md (*andnot<mode>3,
1234 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
1235 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
1236 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
1237 not applied use empty suffix even for TARGET_AVX512VL.
1238 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
1239 is applied, supply evex,evex or evex,evex,evex instead of just
1242 2017-10-20 Julia Koval <julia.koval@intel.com>
1244 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
1245 (OPTION_MASK_ISA_GFNI_UNSET): New.
1246 (ix86_handle_option): Handle OPT_mgfni.
1247 * config/i386/cpuid.h (bit_GFNI): New.
1248 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
1249 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
1250 * config/i386/i386.c (ix86_target_string): Add -mgfni.
1251 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
1252 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
1253 * config/i386/i386.opt: Add mgfni.
1255 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
1257 * config/msp430/msp430.c (msp430_option_override): Disable
1258 -fdelete-null-pointer-checks.
1259 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
1261 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1263 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
1264 of x87 and SSE instructions.
1266 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1268 * asan.c (create_cond_insert_point): Do not update edge count.
1269 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
1270 (afdo_propagate_circuit): Likewise.
1271 (afdo_calculate_branch_prob): Likewise.
1272 (afdo_annotate_cfg): Likewise.
1273 * basic-block.h (struct edge_def): Remove count.
1274 (edge_def::count): New accessor.
1275 * bb-reorder.c (rotate_loop): Update.
1276 (find_traces_1_round): Update.
1277 (connect_traces): Update.
1278 (sanitize_hot_paths): Update.
1279 * cfg.c (unchecked_make_edge): Update.
1280 (make_single_succ_edge): Update.
1281 (check_bb_profile): Update.
1282 (dump_edge_info): Update.
1283 (update_bb_profile_for_threading): Update.
1284 (scale_bbs_frequencies_int): Update.
1285 (scale_bbs_frequencies_gcov_type): Update.
1286 (scale_bbs_frequencies_profile_count): Update.
1287 (scale_bbs_frequencies): Update.
1288 * cfganal.c (connect_infinite_loops_to_exit): Update.
1289 * cfgbuild.c (compute_outgoing_frequencies): Update.
1290 (find_many_sub_basic_blocks): Update.
1291 * cfgcleanup.c (try_forward_edges): Update.
1292 (try_crossjump_to_edge): Update
1293 * cfgexpand.c (expand_gimple_cond): Update
1294 (expand_gimple_tailcall): Update
1295 (construct_exit_block): Update
1296 * cfghooks.c (verify_flow_info): Update
1297 (redirect_edge_succ_nodup): Update
1298 (split_edge): Update
1299 (make_forwarder_block): Update
1300 (duplicate_block): Update
1301 (account_profile_record): Update
1302 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
1303 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
1304 * cfgloopmanip.c (scale_loop_profile): Update.
1306 (lv_adjust_loop_entry_edge): Update.
1307 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
1308 (force_nonfallthru_and_redirect): Update.
1309 (purge_dead_edges): Update.
1310 (rtl_flow_call_edges_add): Update.
1311 * cgraphunit.c (init_lowered_empty_function): Update.
1312 (cgraph_node::expand_thunk): Update.
1313 * gimple-pretty-print.c (dump_probability): Update.
1314 (dump_edge_probability): Update.
1315 * gimple-ssa-isolate-paths.c (isolate_path): Update.
1316 * haifa-sched.c (sched_create_recovery_edges): Update.
1317 * hsa-gen.c (convert_switch_statements): Update.
1318 * ifcvt.c (dead_or_predicable): Update.
1319 * ipa-inline-transform.c (inline_transform): Update.
1320 * ipa-split.c (split_function): Update.
1321 * ipa-utils.c (ipa_merge_profiles): Update.
1322 * loop-doloop.c (add_test): Update.
1323 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
1324 * lto-streamer-in.c (input_cfg): Update.
1325 (input_function): Update.
1326 * lto-streamer-out.c (output_cfg): Update.
1327 * modulo-sched.c (sms_schedule): Update.
1328 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
1329 * predict.c (maybe_hot_edge_p): Update.
1330 (unlikely_executed_edge_p): Update.
1331 (probably_never_executed_edge_p): Update.
1332 (dump_prediction): Update.
1333 (drop_profile): Update.
1334 (propagate_unlikely_bbs_forward): Update.
1335 (determine_unlikely_bbs): Update.
1336 (force_edge_cold): Update.
1337 * profile.c (compute_branch_probabilities): Update.
1338 * reg-stack.c (better_edge): Update.
1339 * shrink-wrap.c (handle_simple_exit): Update.
1340 * tracer.c (better_p): Update.
1341 * trans-mem.c (expand_transaction): Update.
1342 (split_bb_make_tm_edge): Update.
1343 * tree-call-cdce.c: Update.
1344 * tree-cfg.c (gimple_find_sub_bbs): Update.
1345 (gimple_split_edge): Update.
1346 (gimple_duplicate_sese_region): Update.
1347 (gimple_duplicate_sese_tail): Update.
1348 (gimple_flow_call_edges_add): Update.
1349 (insert_cond_bb): Update.
1350 (execute_fixup_cfg): Update.
1351 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
1352 * tree-complex.c (expand_complex_div_wide): Update.
1353 * tree-eh.c (lower_resx): Update.
1354 (unsplit_eh): Update.
1355 (cleanup_empty_eh_move_lp): Update.
1356 * tree-inline.c (copy_edges_for_bb): Update.
1357 (freqs_to_counts): Update.
1358 (copy_cfg_body): Update.
1359 * tree-ssa-dce.c (remove_dead_stmt): Update.
1360 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
1361 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
1362 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
1363 (unloop_loops): Update.
1364 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
1365 * tree-ssa-loop-split.c (connect_loops): Update.
1366 (split_loop): Update.
1367 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
1368 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
1369 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
1370 * tree-ssa-reassoc.c (branch_fixup): Update.
1371 * tree-ssa-tail-merge.c (replace_block_by): Update.
1372 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
1373 (compute_path_counts): Update.
1374 (update_profile): Update.
1375 (recompute_probabilities): Update.
1376 (update_joiner_offpath_counts): Update.
1377 (estimated_freqs_path): Update.
1378 (freqs_to_counts_path): Update.
1379 (clear_counts_path): Update.
1380 (ssa_fix_duplicate_block_edges): Update.
1381 (duplicate_thread_path): Update.
1382 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
1383 (case_bit_test_cmp): Update.
1384 (collect_switch_conv_info): Update.
1385 (gen_inbound_check): Update.
1386 (do_jump_if_equal): Update.
1387 (emit_cmp_and_jump_insns): Update.
1388 * tree-tailcall.c (decrease_profile): Update.
1389 (eliminate_tail_call): Update.
1390 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
1391 (vect_do_peeling): Update.
1392 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
1393 * ubsan.c (ubsan_expand_null_ifn): Update.
1394 (ubsan_expand_ptr_ifn): Update.
1395 * value-prof.c (gimple_divmod_fixed_value): Update.
1396 (gimple_mod_pow2): Update.
1397 (gimple_mod_subtract): Update.
1398 (gimple_ic): Update.
1399 (gimple_stringop_fixed_value): Update.
1401 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1404 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
1406 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
1408 PR rtl-optimization/82395
1409 * ira-color.c (allocno_priority_compare_func): Fix comparison step
1410 based on non_spilled_static_chain_regno_p.
1412 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1414 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
1415 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
1416 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
1418 2017-10-19 Martin Sebor <msebor@redhat.com>
1420 PR tree-optimization/82596
1421 * tree.c (array_at_struct_end_p): Handle STRING_CST.
1423 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1425 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
1426 (get_mem_refs_of_builtin_call): Likewise.
1427 * builtins.c (expand_builtin_apply): Adjust call to
1428 allocate_dynamic_stack_space.
1429 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
1430 the third argument to allocate_dynamic_stack_space, otherwise -1.
1431 (expand_builtin): Deal with all alloca variants.
1432 (is_inexpensive_builtin): Likewise.
1433 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
1434 * calls.c (special_function_p): Deal with all alloca variants.
1435 (initialize_argument_information): Adjust call to
1436 allocate_dynamic_stack_space.
1437 (expand_call): Likewise.
1438 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
1439 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
1440 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
1441 use it for the stack usage computation.
1442 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
1443 * function.c (gimplify_parameters): Call build_alloca_call_expr.
1444 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
1445 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
1446 (in_loop_p): Remove first argument and useless check.
1447 (pass_walloca::execute): Remove useless test and adjust call to above.
1448 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
1449 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
1450 (gimplify_call_expr): Deal with all alloca variants.
1451 * hsa-gen.c (gen_hsa_alloca): Likewise.
1452 (gen_hsa_insns_for_call): Likewise.
1453 * ipa-pure-const.c (special_builtin_state): Likewise.
1454 * tree-chkp.c (chkp_build_returned_bound): Likewise.
1455 * tree-object-size.c (alloc_object_size): Likewise.
1456 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1457 (call_may_clobber_ref_p_1): Likewise.
1458 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
1459 (ccp_fold_stmt): Likewise.
1460 (optimize_stack_restore): Likewise.
1461 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1462 (mark_all_reaching_defs_necessary_1): Likewise.
1463 (propagate_necessity): Likewise.
1464 (eliminate_unnecessary_stmts): Likewise.
1465 * tree.c (build_common_builtin_nodes): Build
1466 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
1467 (build_alloca_call_expr): New function.
1468 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
1469 (CASE_BUILT_IN_ALLOCA): Likewise.
1470 (build_alloca_call_expr): Declare.
1471 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
1473 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1476 * dwarf2out.c (new_die_raw): New static inline function.
1477 (new_die): Use it to create the DIE.
1478 (add_AT_external_die_ref): Likewise.
1479 (clone_die): Likewise.
1480 (clone_as_declaration): Likewise.
1481 (dwarf2out_vms_debug_main_pointer): Likewise.
1482 (base_type_die): Likewise. Remove early return for corner cases.
1483 Do not call add_pubtype on the DIE here.
1484 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
1485 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
1486 typedefs for base types with DW_AT_endianity. Make sure a DIE with
1487 native order exists for base types, attach the DIE manually and call
1488 add_pubtype on it. Do not equate a reverse order DIE to the type.
1490 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
1492 * config/arm/arm.c (align_ok_ldrd_strd): New function.
1493 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
1495 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
1497 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1499 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
1500 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
1501 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
1502 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
1503 * opts.c (sanitizer_opts): Add builtin.
1504 * ubsan.c (instrument_builtin): New function.
1505 (pass_ubsan::execute): Call it.
1506 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
1507 * doc/invoke.texi: Document -fsanitize=builtin.
1509 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
1510 builtins, store max (log2 (align), 0) into uchar field instead of
1511 align into uptr field.
1512 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
1513 store uchar 0 field instead of uptr 0 field.
1514 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
1515 instead of passing one address of struct with 2 locations pass
1516 two addresses of structs with 1 location each.
1517 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
1518 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
1519 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
1520 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
1521 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
1522 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
1523 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
1524 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
1526 2017-10-19 Martin Liska <mliska@suse.cz>
1529 * file-find.c (remove_prefix): Remove.
1530 * file-find.h (remove_prefix): Likewise.
1531 * gcc-ar.c: Remove smartness of lookup.
1533 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1535 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
1536 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
1537 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
1539 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1542 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
1543 (setcc + and to xor + setcc): New peephole2.
1545 2017-10-19 Tom de Vries <tom@codesourcery.com>
1547 * doc/sourcebuild.texi (Test Directives, Variants of
1548 dg-require-support): Add dg-require-stack-size.
1550 2017-10-19 Martin Liska <mliska@suse.cz>
1553 * gimplify.c (gimplify_decl_expr): Do not instrument variables
1554 that have a large alignment.
1555 (gimplify_target_expr): Likewise.
1557 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
1559 PR rtl-optimization/82602
1560 * ira.c (rtx_moveable_p): Return false for volatile asm.
1562 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
1565 * config/i386/i386-modes.def (CCGZ): New CC mode.
1566 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
1567 * config/i386/predicates.md (ix86_comparison_operator):
1569 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1570 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
1571 with double-word subtraction.
1572 (put_condition_code): Handle CCGZmode.
1574 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
1576 * wide-int.cc (debug (const wide_int &)): New.
1577 (debug (const wide_int *)): New.
1578 (debug (const widest_int &)): New.
1579 (debug (const widest_int *)): New.
1581 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
1584 * lra-constraints.c (curr_insn_transform): Use non-input operand
1585 instead of output one for matched reload.
1587 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1589 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
1590 (tree-ssa-loop-ivopts.h): New header file.
1591 (struct builtin_info): New fields.
1592 (classify_builtin_1): Compute and record base and offset parts for
1593 memset builtin partition by calling strip_offset.
1594 (offset_cmp, fuse_memset_builtins): New functions.
1595 (finalize_partitions): Fuse adjacent memset partitions by calling
1597 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
1598 Expose the interface.
1599 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
1601 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1603 PR tree-optimization/82574
1604 * tree-loop-distribution.c (find_single_drs): New parameter. Check
1605 that data reference must be executed exactly once per iteration
1606 against the outermost loop in nest.
1607 (classify_partition): Update call to above function.
1609 2017-10-18 Richard Biener <rguenther@suse.de>
1611 PR tree-optimization/82591
1612 * graphite.c (graphite_transform_loops): Move code gen message
1614 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1615 Here. Handle scop_to_isl_ast failing.
1616 (scop_to_isl_ast): Limit the number of ISL operations.
1618 2017-10-18 Richard Biener <rguenther@suse.de>
1620 * graphite-isl-ast-to-gimple.c
1621 (translate_isl_ast_to_gimple::set_rename): Simplify.
1622 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
1623 (graphite_copy_stmts_from_block): ... here.
1624 (copy_bb_and_scalar_dependences): Simplify.
1625 (add_parameters_to_ivs_params): Canonicalize.
1626 (generate_entry_out_of_ssa_copies): Simplify.
1627 * graphite-sese-to-poly.c (extract_affine_name): Simplify
1628 by passing in ISL dimension.
1629 (parameter_index_in_region_1): Rename to ...
1630 (parameter_index_in_region): ... this.
1631 (extract_affine): Adjust assert, pass down parameter index.
1632 (add_param_constraints): Use range-info when available.
1633 (build_scop_context): Adjust.
1634 * sese.c (new_sese_info): Adjust.
1635 (free_sese_info): Likewise.
1636 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
1637 Remove unused typedefs.
1638 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
1640 2017-10-18 Martin Liska <mliska@suse.cz>
1642 * combine.c (simplify_compare_const): Add gcc_fallthrough.
1644 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1646 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
1647 (s390_sched_init): Do not reset s390_sched_state if we entered the
1648 current basic block via a fallthru edge and all others are unlikely.
1650 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1652 * config/s390/s390.c (NUM_SIDES): New variable.
1653 (LONGRUNNING_THRESHOLD): New variable.
1654 (LATENCY_FACTOR): New variable.
1655 (s390_sched_score): Decrease score for long-running instructions on
1657 (s390_sched_variable_issue): Perform bookkeeping for long-running
1660 2017-10-18 Richard Biener <rguenther@suse.de>
1662 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
1663 Simplify with removal of the parameter rename map.
1664 (set_rename): Likewise.
1665 (should_copy_to_new_region): Likewise.
1666 (graphite_copy_stmts_from_block): Likewise.
1667 (copy_bb_and_scalar_dependences): Remove initialization of
1668 unused copied_bb_map.
1670 (copy_internal_parameters): Likewise.
1671 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
1672 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
1673 Use INTEGRAL_TYPE_P.
1674 (parameter_index_in_region_1): Rename to ...
1675 (assign_parameter_index_in_region): ... this. Assert we have
1676 a parameter we handle.
1677 (scan_tree_for_params): Adjust.
1678 * sese.h (parameter_rename_map_t): Remove.
1679 (struct sese_info_t): Remove unused parameter_rename_map and
1680 copied_bb_map members.
1681 * sese.c (new_sese_info): Adjust.
1682 (free_sese_info): Likewise.
1684 2017-10-18 Martin Liska <mliska@suse.cz>
1687 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
1688 on an abnormal edge.
1690 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
1692 * doc/invoke.texi (ffunction-sections and fdata-sections):
1695 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1697 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
1698 the use statement can throw internally.
1700 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1702 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
1703 any RTX present on the RHS of a SET.
1704 * compare-elim.c (try_eliminate_compare): Restore comment.
1706 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1708 * langhooks.h (struct lang_hooks): Document that tree_size langhook
1709 may be also called on tcc_type nodes.
1710 * langhooks.c (lhd_tree_size): Likewise.
1712 2017-10-17 David Malcolm <dmalcolm@redhat.com>
1714 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
1715 format_warning_at_substring.
1716 (maybe_warn): Convert source_range * param to a location_t. Pass
1717 UNKNOWN_LOCATION rather than NULL to fmtwarn.
1718 (format_directive): Remove code to extract source_ranges and
1719 source_range * in favor of just a location_t.
1720 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
1722 * substring-locations.c (format_warning_va): Convert
1723 source_range * param to a location_t.
1724 (format_warning_at_substring): Likewise.
1725 * substring-locations.h (format_warning_va): Likewise.
1726 (format_warning_at_substring): Likewise.
1728 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
1730 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
1732 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
1733 and scatter/gather ops.
1735 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
1736 vec_gather_load and vec_scatter_store.
1737 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
1738 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
1739 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
1740 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
1741 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
1742 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
1744 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
1746 * reg-stack.c (compare_for_stack_reg): Add bool argument.
1747 Detect FTST instruction and handle its register pops. Only pop
1748 second operand if can_pop_second_op is true.
1749 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
1750 set can_pop_second_op to false in the compare_for_stack_reg call.
1752 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
1753 output_fp_compare for stack register operands.
1754 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
1755 instructions here. Do not emit stack register pops here. Assert
1756 that FCOMPP pops next to top stack register. Rewrite function.
1758 2017-10-17 Nathan Sidwell <nathan@acm.org>
1761 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
1762 use DECL_ASSEMBLER_NAME_RAW.
1765 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
1768 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
1769 Wilco Dijkstra <wilco.dijkstra@arm.com>
1771 * builtins.c (expand_builtin_update_setjmp_buf): Add a
1772 converstion to Pmode from the buf_addr.
1774 2017-10-17 Richard Biener <rguenther@suse.de>
1776 * graphite-dependences.c (scop_get_reads_and_writes): Change
1777 output parameters to references.
1779 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
1781 PR 71026/tree-optimization
1782 * fold-const.c (distribute_real_division): Removed.
1783 (fold_binary_loc): Remove calls to distribute_real_divison.
1785 2017-10-17 Richard Biener <rguenther@suse.de>
1787 * graphite-scop-detection.c
1788 (scop_detection::stmt_has_simple_data_refs_p): Always use
1789 the full nest as region.
1790 (try_generate_gimple_bb): Likewise.
1791 * sese.c (scalar_evolution_in_region): Simplify now that
1792 SCEV can handle instantiation in regions.
1793 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
1794 in the non-loop part of a function if requested.
1796 2017-10-17 Richard Biener <rguenther@suse.de>
1798 PR tree-optimization/82563
1799 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
1801 (graphite_regenerate_ast_isl): Call it.
1802 * graphite-scop-detection.c (build_scops): Remove entry edge split.
1804 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1806 PR tree-optimization/82549
1807 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
1808 Formatting fixes. Instead of calling make_bit_field_ref with negative
1811 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
1813 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
1814 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
1815 _mm_maskz_reduce_ss): New.
1816 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
1817 __builtin_ia32_reducess_mask): Ditto..
1818 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
1819 * config/i386/sse.md (reduces<mode>): Renamed to ...
1820 (reduces<mode><mask_scalar_name>): ... this.
1821 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
1823 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
1824 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
1826 2017-10-16 David Malcolm <dmalcolm@redhat.com>
1828 * Makefile.in (OBJS): Add unique-ptr-tests.o.
1829 * selftest-run-tests.c (selftest::run_tests): Call
1830 selftest::unique_ptr_tests_cc_tests.
1831 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
1832 * unique-ptr-tests.cc: New file.
1834 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
1837 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1839 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1840 (make_hard_regno_born, make_hard_regno_dead): Update
1841 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
1843 2017-10-16 Jeff Law <law@redhat.com>
1845 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1847 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
1849 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
1851 2017-10-16 Olivier Hainque <hainque@adacore.com>
1853 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
1854 with_cpu if we were configured for an e500v2 target cpu name.
1856 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
1858 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
1859 * doc/invoke.texi: Document +nodsp as a valid extension for
1862 2017-10-16 Martin Liska <mliska@suse.cz>
1864 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
1865 (test_set_range): Likewise.
1866 (test_range_functions): Rename to ...
1867 (test_bit_in_range): ... this.
1868 (sbitmap_c_tests): Add new test.
1870 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1872 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
1874 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
1875 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
1877 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1879 * config/aarch64/aarch64-builtins.c
1880 (aarch64_types_quadopu_lane_qualifiers): New.
1881 (TYPES_QUADOPU_LANE): New.
1882 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
1883 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
1884 (aarch64_<sur>dot_laneq<vsi2qi>): New.
1885 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
1886 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
1887 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
1888 (Vdottype, DOTPROD): New.
1889 (sur): Add SDOT and UDOT.
1891 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1893 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
1894 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
1895 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
1897 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
1898 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
1899 Enable TARGET_DOTPROD.
1900 (cortex-a75.cortex-a55): Likewise.
1901 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
1903 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1905 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
1906 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
1908 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
1910 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
1911 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
1912 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
1913 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
1914 * config/arm/types.md (neon_dot, neon_dot_q): New.
1915 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
1917 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1919 * config/arm/arm.h (TARGET_DOTPROD): New.
1920 * config/arm/arm.c (arm_arch_dotprod): New.
1921 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
1922 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
1923 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
1924 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
1925 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
1926 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
1927 * doc/invoke.texi (armv8.2-a): Document dotprod
1929 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
1931 * i386.c (ix86_vec_cost): New function.
1932 (ix86_rtx_costs): Handle vector operations better.
1933 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
1934 * x86-tune-costs.h: Add new costs to all tables.
1936 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
1938 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
1940 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
1941 divsd, sqrtss and sqrtsd
1942 * x86-tune-costs.h: Add new entries to all costs.
1943 (znver1_cost): Fix to match real instruction latencies.
1945 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1946 Michael Collison <michael.collison@arm.com>
1948 * compare-elim.c: Include emit-rtl.h.
1949 (can_merge_compare_into_arith): New function.
1950 (try_validate_parallel): Likewise.
1951 (try_merge_compare): Likewise.
1952 (try_eliminate_compare): Call the above when no previous clobber
1954 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
1957 2017-10-14 Jakub Jelinek <jakub@redhat.com>
1961 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
1962 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
1966 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
1967 to be any operand_equal_p operands. For & (B - 1) require
1968 B to be power of 2. Recognize
1969 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
1971 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1974 * optabs.c (expand_memory_blockage): Fix call of
1975 targetm.have_memory_blockage.
1977 2017-10-14 Jakub Jelinek <jakub@redhat.com>
1980 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
1981 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
1982 objects to extra_objs instead of overwriting it.
1984 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1986 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
1987 Use any_fp_register_operand as operand[3] predicate. Simplify
1988 equality test for operands[2] and operands[4] memory location.
1989 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
1990 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
1991 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
1992 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
1993 any_fp_register_operand as operand[1] predicate. Simplify
1994 equality test for operands[0] and operands[3] memory location.
1995 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
1996 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
1997 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
1999 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2001 * target-insns.def: Add memory_blockage.
2002 * optabs.c (expand_memory_blockage): New function.
2003 (expand_asm_memory_barrier): Rename ...
2004 (expand_asm_memory_blockage): ... to this.
2005 (expand_mem_thread_fence): Call expand_memory_blockage
2006 instead of expand_asm_memory_barrier.
2007 (expand_mem_singnal_fence): Ditto.
2008 (expand_atomic_load): Ditto.
2009 (expand_atomic_store): Ditto.
2010 * doc/md.texi (Standard Pattern Names For Generation):
2011 Document memory_blockage instruction pattern.
2013 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
2015 * config/rl78/rl78.c (rl78_emit_libcall): New function.
2016 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
2017 * config/rl78/rl78.md: New define_expand "adddi3".
2019 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2021 * cfghooks.c (verify_flow_info): Disable check that all probabilities
2024 2017-10-13 Jeff Law <law@redhat.com>
2026 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
2028 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2031 * internal-fn.c (expand_mul_overflow): If both operands have
2032 the same highpart of -1 or 0 and the topmost bit of lowpart
2033 is different, overflow is if res <= 0 rather than res < 0.
2035 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
2037 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
2038 TARGET_P9_VECTOR code for unaligned_load case.
2040 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2042 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
2044 2017-10-13 Nathan Sidwell <nathan@acm.org>
2046 * tree-core.h (tree_contains_struct): Make bool.
2047 * tree.c (tree_contains_struct): Likewise.
2048 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
2049 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
2050 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
2051 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
2052 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
2054 2017-10-13 Richard Biener <rguenther@suse.de>
2056 * graphite-isl-ast-to-gimple.c
2057 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
2058 parameters and dominance check.
2059 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
2060 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
2061 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
2062 Do not update SSA form here or do intermediate IL verification.
2063 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
2064 (graphite_initialize): Remove check on the number of loops in
2065 the function and inline into graphite_transform_loops.
2066 (graphite_finalize): Inline into graphite_transform_loops.
2067 (graphite_transform_loops): Perform SSA update and IL verification
2069 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
2071 2017-10-13 Richard Biener <rguenther@suse.de>
2073 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
2074 graphite_expression_type_precision): Avoid global constructor
2076 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
2077 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
2078 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
2079 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
2080 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
2081 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
2083 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
2086 * config/i386/i386.h (ix86_red_zone_size): New.
2087 * config/i386/i386.md (push peephole2s): Replace
2088 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
2090 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2091 Alan Hayward <alan.hayward@arm.com>
2092 David Sherwood <david.sherwood@arm.com>
2094 * combine.c (can_change_dest_mode): Reject changes in
2095 REGMODE_NATURAL_SIZE.
2097 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2098 Alan Hayward <alan.hayward@arm.com>
2099 David Sherwood <david.sherwood@arm.com>
2101 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
2102 (expand_debug_source_expr): Likewise.
2103 * combine.c (combine_simplify_rtx): Likewise.
2104 * cse.c (fold_rtx): Likewise.
2105 * fwprop.c (canonicalize_address): Likewise.
2106 * targhooks.c (default_shift_truncation_mask): Likewise.
2108 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2109 Alan Hayward <alan.hayward@arm.com>
2110 David Sherwood <david.sherwood@arm.com>
2112 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
2113 (widened_mode): Likewise.
2114 (expand_unop): Likewise.
2115 * ree.c (transform_ifelse): Likewise.
2116 (merge_def_and_ext): Likewise.
2117 (combine_reaching_defs): Likewise.
2118 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2120 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2121 Alan Hayward <alan.hayward@arm.com>
2122 David Sherwood <david.sherwood@arm.com>
2124 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
2125 * combine.c (gen_lowpart_for_combine): Likewise.
2126 * dwarf2out.c (rtl_for_decl_location): Likewise.
2127 * final.c (alter_subreg): Likewise.
2128 * rtlhooks.c (gen_lowpart_general): Likewise.
2129 (gen_lowpart_if_possible): Likewise.
2131 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2132 Alan Hayward <alan.hayward@arm.com>
2133 David Sherwood <david.sherwood@arm.com>
2135 * calls.c (expand_call): Use subreg_lowpart_offset.
2136 * cse.c (cse_insn): Likewise.
2137 * regcprop.c (copy_value): Likewise.
2138 (copyprop_hardreg_forward_1): Likewise.
2140 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2143 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
2144 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
2145 =Q constraints to +Q and into insn condition add check
2146 that operands[0] and operands[1] are equal.
2147 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
2148 =Q constraints to +Q and into insn condition add check
2149 that operands[0] is equal to either operands[1] or operands[2].
2152 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
2153 instead of handling MINUS_EXPR twice (once for each argument),
2154 canonicalize operand order and handle just once, use rtype where
2155 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
2158 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
2159 any values of __C while still being pattern recognizable as a simple
2162 2017-10-13 Richard Biener <rguenther@suse.de>
2164 PR tree-optimization/82451
2166 2017-10-02 Richard Biener <rguenther@suse.de>
2168 PR tree-optimization/82355
2169 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2170 a mapping for the enclosing loop but avoid generating one for
2172 (copy_bb_and_scalar_dependences): Remove premature codegen
2173 error on PHIs in blocks duplicated into multiple places.
2174 * graphite-scop-detection.c
2175 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2176 in the region use it as loop and nest to analyze the DR in.
2177 (try_generate_gimple_bb): Likewise.
2178 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2179 (add_loop_constraints): For blocks in a loop not in the region
2180 create a dimension with a single iteration.
2181 * sese.h (gbb_loop_at_index): Remove assert.
2183 * cfgloop.c (loop_preheader_edge): For the loop tree root
2184 return the single successor of the entry block.
2185 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2186 Reset the SCEV hashtable and niters.
2187 * graphite-scop-detection.c
2188 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
2189 assert that we only have POLYNOMIAL_CHREC that vary in loops
2190 contained in the region.
2191 (scop_detection::graphite_can_represent_expr): Adjust.
2192 (scop_detection::stmt_has_simple_data_refs_p): For loops
2193 not in the region set loop to NULL. The nest is now the
2194 entry edge to the region.
2195 (try_generate_gimple_bb): Likewise.
2196 * sese.c (scalar_evolution_in_region): Adjust for
2197 instantiate_scev change.
2198 * tree-data-ref.h (graphite_find_data_references_in_stmt):
2199 Make nest parameter the edge into the region.
2200 (create_data_ref): Likewise.
2201 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
2202 entry edge into a region and adjust instantiate_scev calls.
2203 (create_data_ref): Likewise.
2204 (graphite_find_data_references_in_stmt): Likewise.
2205 (find_data_references_in_stmt): Pass the loop preheader edge
2206 from the nest argument.
2207 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
2208 parameter the edge into the region.
2209 (instantiate_parameters): Use the loop preheader edge as entry.
2210 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
2212 (get_instantiated_value_entry): Make instantiate_below parameter
2213 the edge into the region.
2214 (instantiate_scev_name): Likewise. Adjust dominance checks,
2215 when we cannot use loop-based instantiation instantiate by
2216 walking use-def chains.
2217 (instantiate_scev_poly): Adjust.
2218 (instantiate_scev_binary): Likewise.
2219 (instantiate_scev_convert): Likewise.
2220 (instantiate_scev_not): Likewise.
2221 (instantiate_array_ref): Remove.
2222 (instantiate_scev_3): Likewise.
2223 (instantiate_scev_2): Likewise.
2224 (instantiate_scev_1): Likewise.
2225 (instantiate_scev_r): Do not blindly handle N-operand trees.
2226 Do not instantiate array-refs. Handle all constants and invariants.
2227 (instantiate_scev): Make instantiate_below parameter
2228 the edge into the region.
2229 (resolve_mixers): Use the loop preheader edge for the region
2230 parameter to instantiate_scev_r.
2231 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
2233 2017-10-13 Richard Biener <rguenther@suse.de>
2235 PR tree-optimization/82525
2236 * graphite-isl-ast-to-gimple.c
2237 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
2239 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
2240 Fail code generation when we cannot represent the isl integer.
2241 (binary_op_to_tree): Elide modulo operations that are no-ops
2242 in the type we code generate. Remove now superfluous code
2245 2017-10-13 Richard Biener <rguenther@suse.de>
2247 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
2248 (scop_detection::harmful_loop_in_region): Remove premature
2249 IV type restriction.
2250 (scop_detection::graphite_can_represent_scev): We can handle
2251 pointer IVs just fine.
2253 2017-10-13 Alan Modra <amodra@gmail.com>
2255 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
2256 "Clobbers and Scratch Registers". Add paragraph on
2257 alternative to clobbers for scratch registers and OpenBLAS
2260 2017-10-13 Alan Modra <amodra@gmail.com>
2262 * doc/extend.texi (Clobbers): Correct vax example. Delete old
2263 example of a memory input for a string of known length. Move
2264 commentary out of table. Add a number of new examples
2265 covering array memory inputs.
2267 2017-10-12 Martin Liska <mliska@suse.cz>
2269 PR tree-optimization/82493
2270 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
2271 (test_range_functions): New function.
2272 (sbitmap_c_tests): Likewise.
2273 * selftest-run-tests.c (selftest::run_tests): Run new tests.
2274 * selftest.h (sbitmap_c_tests): New function.
2276 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
2278 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
2280 * config/rs6000/amo.h: Fix spacing issue.
2282 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2285 * config/i386/i386.md (*ashl<mode>3_mask_1,
2286 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
2287 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
2290 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
2292 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
2293 (profile_probability): Set max_probability
2294 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
2296 * profile-count.c (profile_probability::differs_from_p): Do not
2297 rely on max_probaiblity == 10000
2299 2017-10-12 Jeff Law <law@redhat.com>
2301 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
2304 2017-10-12 Martin Sebor <msebor@redhat.com>
2308 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
2309 (handle_alias_pairs): Call it.
2310 * common.opt (-Wattribute-alias): New option.
2311 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
2312 * doc/invoke.texi (-Wattribute-alias): Document.
2314 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
2317 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2319 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2321 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2322 (make_hard_regno_born, make_hard_regno_dead): Update
2323 bb_killed_pseudos and bb_gen_pseudos.
2325 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
2327 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
2329 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
2331 * config/alpha/alpha.c (alpha_split_conditional_move):
2332 Use std::swap instead of manually swapping.
2333 (alpha_stdarg_optimize_hook): Ditto.
2334 (alpha_canonicalize_comparison): Ditto.
2336 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2338 * tree-loop-distribution.c (struct builtin_info): New struct.
2339 (struct partition): Refactor fields into struct builtin_info.
2340 (partition_free): Free struct builtin_info.
2341 (build_size_arg_loc, build_addr_arg_loc): Delete.
2342 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
2343 information from struct builtin_info.
2344 (find_single_drs): New function refactored from classify_partition.
2345 Also moved builtin validity checks to this function.
2346 (compute_access_range, alloc_builtin): New functions.
2347 (classify_builtin_st, classify_builtin_ldst): New functions.
2348 (classify_partition): Refactor code into functions find_single_drs,
2349 classify_builtin_st and classify_builtin_ldst.
2350 (distribute_loop): Don't do runtime alias check when distributing
2352 (find_seed_stmts_for_distribution): New function.
2353 (pass_loop_distribution::execute): Refactor code finding seed
2354 stmts into above function. Support distribution for the innermost
2355 two-level loop nest. Adjust dump information.
2357 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2359 * tree-loop-distribution.c: Adjust the general comment.
2360 (NUM_PARTITION_THRESHOLD): New macro.
2361 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
2362 (classify_partition): Skip builtin pattern of loop nest's inner loop.
2363 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
2364 in call to build_partition_graph.
2365 (finalize_partitions): New parameter. Make loop distribution more
2366 conservative by fusing more partitions.
2367 (distribute_loop): Don't do runtime alias check in case of loop nest
2369 (find_seed_stmts_for_distribution): New function.
2370 (prepare_perfect_loop_nest): New function.
2371 (pass_loop_distribution::execute): Refactor code finding seed stmts
2372 and loop nest into above functions. Support loop nest distribution.
2373 Adjust dump information accordingly.
2375 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2377 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
2378 and set PTYPE_SEQUENTIAL for merged partition.
2380 2017-10-12 Richard Biener <rguenther@suse.de>
2382 PR tree-optimization/69728
2384 2017-09-19 Richard Biener <rguenther@suse.de>
2386 PR tree-optimization/69728
2387 * graphite-sese-to-poly.c (schedule_error): New global.
2388 (add_loop_schedule): Handle empty domain by failing the
2390 (build_original_schedule): Handle schedule_error.
2392 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
2393 domain by returning an unchanged schedule.
2395 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2397 * genrecog.c (validate_pattern): For VEC_SELECT verify that
2398 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
2400 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
2402 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
2405 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2408 * expr.c (store_field): Don't optimize away bitsize == 0 store
2409 from CALL_EXPR with addressable return type.
2411 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2413 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
2414 * config/rs6000/rs6000.md (sel): Delete mode attribute.
2415 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
2416 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
2417 TARGET_ISEL instead of TARGET_ISEL<sel>.
2419 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
2421 * config/rs6000/rs6000.c
2422 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
2424 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2426 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
2428 (reg_or_cint_operand): Fix comment.
2429 (reg_or_zero_operand): New predicate.
2430 * config/rs6000/rs6000-protos.h (output_isel): Delete.
2431 * config/rs6000/rs6000.c (output_isel): Delete.
2432 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
2433 instead of reg_or_cint_operand. Output instruction directly (not via
2435 (isel_unsigned_<mode>): Ditto.
2436 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
2437 gpc_reg_operand. Add an instruction alternative for this. Output
2438 instruction directly.
2439 (*isel_reversed_unsigned_<mode>): Ditto.
2441 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2443 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
2444 (TARGET_CANONICALIZE_COMPARISON): Define.
2446 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
2449 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
2450 Check whether the dest is REG before adding REG_EQUIV note.
2452 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2455 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2457 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2458 (make_hard_regno_born, make_hard_regno_dead): Update
2459 bb_killed_pseudos and bb_gen_pseudos.
2461 2017-10-11 Nathan Sidwell <nathan@acm.org>
2463 * incpath.h (enum incpath_kind): Name enum, prefix values.
2464 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
2465 * incpath.c (heads, tails): Use INC_MAX.
2466 (add_env_var_paths, add_standard_paths): Use incpath_kind.
2467 (merge_include_chains, split_quote_chain,
2468 register_include_chains): Update incpath_kind names.
2469 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
2470 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
2472 (add_framework_path, darwin_register_objc_includes): Likewise.
2473 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
2475 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2477 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
2478 Do not use float_operator operator predicate.
2479 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2480 * config/i386/predicates.md (float_operator): Remove predicate.
2482 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2484 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
2485 (*jccxf_i387): Ditto.
2486 (*jcc<mode>_i387): Ditto.
2487 (*jccu<mode>_i387): Ditto.
2488 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
2489 (*jcc_*_i387 splitters): Remove.
2490 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
2491 * config/i386/i386.c (ix86_split_fp_branch): Remove.
2492 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
2495 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2497 * profile-count.h (slow_safe_scale_64bit): New function.
2498 (safe_scale_64bit): New inline.
2499 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
2500 * profile-count.c: Include wide-int.h
2501 (slow_safe_scale_64bit): New.
2503 2017-10-11 Nathan Sidwell <nathan@acm.org>
2505 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
2506 HAS_DECL_ASSEMBLER_NAME_P.
2507 * gimple-expr.c (gimple_decl_printable_name: Check
2508 HAS_DECL_ASSEMBLER_NAME_P too.
2509 * ipa-utils.h (type_in_anonymous_namespace_p): Check
2510 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
2511 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
2512 * passes.c (rest_of_decl_compilation): Check
2513 HAS_DECL_ASSEMBLER_NAME_P too.
2514 * recog.c (verify_changes): Likewise.
2515 * tree-pretty-print.c (dump_decl_name): Likewise.
2516 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
2518 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
2519 (SET_DECL_ASSEMBLER_NAME): Use it.
2520 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
2521 (COPY_DECL_ASSEMBLER_NAME): Likewise.
2522 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
2524 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2526 * config.gcc (i386, x86_64): Add extra objects.
2527 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
2528 (ix86_min_insn_size): Declare.
2529 (ix86_issue_rate): Declare.
2530 (ix86_adjust_cost): Declare.
2531 (ia32_multipass_dfa_lookahead): Declare.
2532 (ix86_macro_fusion_p): Declare.
2533 (ix86_macro_fusion_pair_p): Declare.
2534 (ix86_bd_has_dispatch): Declare.
2535 (ix86_bd_do_dispatch): Declare.
2536 (ix86_core2i7_init_hooks): Declare.
2537 (ix86_atom_sched_reorder): Declare.
2538 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
2539 (COSTS_N_BYTES): Move to x86-tune-costs.h.
2540 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
2541 (rip_relative_addr_p): Rename to ...
2542 (ix86_rip_relative_addr_p): ... this one; export.
2543 (memory_address_length): Update.
2544 (ix86_issue_rate): Move to x86-tune-sched.c.
2545 (ix86_flags_dependent): Move to x86-tune-sched.c.
2546 (ix86_agi_dependent): Move to x86-tune-sched.c.
2547 (exact_dependency_1): Move to x86-tune-sched.c.
2548 (exact_store_load_dependency): Move to x86-tune-sched.c.
2549 (ix86_adjust_cost): Move to x86-tune-sched.c.
2550 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
2551 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
2552 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
2553 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
2554 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
2555 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
2556 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
2557 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
2558 (min_insn_size): Rename to ...
2559 (ix86_min_insn_size): ... this one; export.
2560 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
2561 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
2562 (core2i7_first_cycle_multipass_backtrack): Move to
2563 x86-tune-sched-core.c.
2564 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
2565 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
2566 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
2567 (ix86_avoid_jump_mispredicts): Update.
2568 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
2569 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
2570 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
2571 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
2572 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
2573 (MAX_INSN): Move to ix86-tune-sched-bd.c.
2574 (MAX_IMM): Move to ix86-tune-sched-bd.c.
2575 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
2576 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
2577 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
2578 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
2579 (MAX_STORE): Move to ix86-tune-sched-bd.c.
2580 (BIG): Move to ix86-tune-sched-bd.c.
2581 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
2582 (enum insn_path): Move to ix86-tune-sched-bd.c.
2583 (get_mem_group): Move to ix86-tune-sched-bd.c.
2584 (is_cmp): Move to ix86-tune-sched-bd.c.
2585 (dispatch_violation): Move to ix86-tune-sched-bd.c.
2586 (is_branch): Move to ix86-tune-sched-bd.c.
2587 (is_prefetch): Move to ix86-tune-sched-bd.c.
2588 (init_window): Move to ix86-tune-sched-bd.c.
2589 (allocate_window): Move to ix86-tune-sched-bd.c.
2590 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
2591 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
2592 (process_end_window): Move to ix86-tune-sched-bd.c.
2593 (allocate_next_window): Move to ix86-tune-sched-bd.c.
2594 (find_constant): Move to ix86-tune-sched-bd.c.
2595 (get_num_immediates): Move to ix86-tune-sched-bd.c.
2596 (has_immediate): Move to ix86-tune-sched-bd.c.
2597 (get_insn_path): Move to ix86-tune-sched-bd.c.
2598 (get_insn_group): Move to ix86-tune-sched-bd.c.
2599 (count_num_restricted): Move to ix86-tune-sched-bd.c.
2600 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
2601 (add_insn_window): Move to ix86-tune-sched-bd.c.
2602 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
2603 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
2604 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
2605 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
2606 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
2607 (do_dispatch): Move to ix86-tune-sched-bd.c.
2608 (has_dispatch): Move to ix86-tune-sched-bd.c.
2609 * config/i386/t-i386: Add new object files.
2610 * config/i386/x86-tune-costs.h: New file.
2611 * config/i386/x86-tune-sched-atom.c: New file.
2612 * config/i386/x86-tune-sched-bd.c: New file.
2613 * config/i386/x86-tune-sched-core.c: New file.
2614 * config/i386/x86-tune-sched.c: New file.
2616 2017-10-11 Liu Hao <lh_mouse@126.com>
2618 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
2619 the generic version below instead.
2620 (should_colorize): Recognize Windows consoles as terminals
2622 * pretty-print.c [__MINGW32__] (write_all): New function.
2623 [__MINGW32__] (find_esc_head): Likewise.
2624 [__MINGW32__] (find_esc_terminator): Likewise.
2625 [__MINGW32__] (eat_esc_sequence): Likewise.
2626 [__MINGW32__] (mingw_ansi_fputs): New function that handles
2628 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
2631 2017-10-11 Richard Biener <rguenther@suse.de>
2633 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
2634 Properly call analyze_scalar_evolution with the loop of the stmt.
2636 2017-10-11 Richard Biener <rguenther@suse.de>
2638 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
2639 * tree-core.h (tree_base): Add chrec_var union member.
2640 * tree.h (CHREC_VAR): Remove.
2641 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
2642 * tree-chrec.h (build_polynomial_chrec): Adjust.
2643 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
2644 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
2646 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
2648 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
2649 * match.pd: ... here.
2650 ((T) X == (T) Y): Relax condition.
2652 2017-10-11 Bin Cheng <bin.cheng@arm.com>
2654 PR tree-optimization/82472
2655 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
2657 (break_alias_scc_partitions): Update postorder number.
2659 2017-10-11 Martin Liska <mliska@suse.cz>
2662 * opts.c (parse_no_sanitize_attribute): Do not use error_value
2664 * opts.h (parse_no_sanitize_attribute): Remove last argument.
2666 2017-10-11 Martin Liska <mliska@suse.cz>
2668 * print-rtl.c (print_insn): Move declaration of idbuf
2669 to same scope as name.
2671 2017-10-11 Martin Liska <mliska@suse.cz>
2676 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2679 2017-10-11 Martin Liska <mliska@suse.cz>
2682 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2685 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
2687 * config.gcc (default_gnu_indirect_function): Default to yes for
2688 arm*-*-linux* with glibc.
2690 2017-10-11 Richard Biener <rguenther@suse.de>
2692 * tree-scalar-evolution.c (get_scalar_evolution): Handle
2693 default-defs and types we do not want to analyze.
2694 (interpret_loop_phi): Replace unreachable code with an assert.
2695 (compute_scalar_evolution_in_loop): Remove and inline ...
2696 (analyze_scalar_evolution_1): ... here, replacing condition with
2697 what makes the intent clearer. Remove handling of cases
2698 get_scalar_evolution now handles.
2700 2017-10-10 Jim Wilson <wilson@tuliptree.org>
2702 PR rtl-optimization/81434
2703 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
2704 comment for main loop. In sched_group_found if, also add checks for
2705 pass and min_cost_group.
2707 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
2709 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
2710 (rs6000_insn_cost): New function.
2711 * config/rs6000/rs6000.md (cost): New attribute.
2713 2017-10-10 Jakub Jelinek <jakub@redhat.com>
2714 H.J. Lu <hongjiu.lu@intel.com>
2718 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
2719 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
2720 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
2721 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
2722 to OPTION_MASK_ISA_AVX512VL - builtins that have both
2723 OPTION_MASK_ISA_MMX and some other bit set require both
2724 mmx and the ISAs without the mmx bit.
2725 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
2726 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
2727 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
2728 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
2729 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
2730 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
2731 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
2732 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
2733 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
2734 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
2735 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
2736 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
2737 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
2738 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
2739 Add OPTION_MASK_ISA_MMX.
2741 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
2743 * config.gcc (armv7*-*-freebsd*): New target.
2744 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
2746 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
2748 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
2749 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
2752 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
2754 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
2755 a template parameter.
2756 (WIDE_INT_REF_FOR): Update accordingly.
2757 * tree.h (wi::int_traits <const_tree>): Delete.
2758 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
2759 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
2760 (wi::tree_to_wide_ref): New typedef.
2761 (wi::to_wide): New function.
2762 * calls.c (get_size_range): Use wi::to_wide when operating on
2764 * cgraph.c (cgraph_node::create_thunk): Likewise.
2765 * config/i386/i386.c (ix86_data_alignment): Likewise.
2766 (ix86_local_alignment): Likewise.
2767 * dbxout.c (stabstr_O): Likewise.
2768 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
2769 * expr.c (const_vector_from_tree): Likewise.
2770 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
2771 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
2772 (fold_negate_expr_1, int_const_binop_1, const_binop)
2773 (fold_convert_const_int_from_real, optimize_bit_field_compare)
2774 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
2775 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
2776 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
2777 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
2778 (fold_not_const, round_up_loc): Likewise.
2779 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
2780 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
2781 (alloca_call_type): Likewise.
2782 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
2783 * godump.c (go_output_typedef): Likewise.
2784 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
2785 * internal-fn.c (get_min_precision): Likewise.
2786 * ipa-cp.c (ipcp_store_vr_results): Likewise.
2787 * ipa-polymorphic-call.c
2788 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
2789 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
2790 (ipa_modify_call_arguments): Likewise.
2791 * match.pd: Likewise.
2792 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
2793 * print-tree.c (print_node_brief, print_node): Likewise.
2794 * stmt.c (expand_case): Likewise.
2795 * stor-layout.c (layout_type): Likewise.
2796 * tree-affine.c (tree_to_aff_combination): Likewise.
2797 * tree-cfg.c (group_case_labels_stmt): Likewise.
2798 * tree-data-ref.c (dr_analyze_indices): Likewise.
2799 (prune_runtime_alias_test_list): Likewise.
2800 * tree-dump.c (dequeue_and_dump): Likewise.
2801 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
2802 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
2803 * tree-pretty-print.c (dump_generic_node): Likewise.
2804 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
2805 (simple_iv_with_niters): Likewise.
2806 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2807 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
2808 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
2809 * tree-ssa-loop-niter.c (split_to_var_and_offset)
2810 (refine_value_range_using_guard, number_of_iterations_ne_max)
2811 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
2812 (get_cst_init_from_scev, record_nonwrapping_iv)
2813 (scev_var_range_cant_overflow): Likewise.
2814 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
2815 * tree-ssa-pre.c (compute_avail): Likewise.
2816 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
2817 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
2818 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
2819 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
2820 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
2821 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
2822 (dump_case_nodes, try_switch_expansion): Likewise.
2823 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
2824 (vect_do_peeling): Likewise.
2825 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
2826 * tree-vect-stmts.c (vectorizable_load): Likewise.
2827 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
2828 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
2829 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
2830 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
2831 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
2832 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
2833 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
2834 (evrp_dom_walker::before_dom_children): Likewise.
2835 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
2836 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
2837 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
2838 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
2839 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
2840 (get_range_pos_neg): Likewise.
2841 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
2842 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
2843 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
2844 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
2845 * config/avr/avr.c (avr_fold_builtin): Likewise.
2846 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
2847 * config/msp430/msp430.c (msp430_attr): Likewise.
2848 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
2849 * config/powerpcspe/powerpcspe-c.c
2850 (altivec_resolve_overloaded_builtin): Likewise.
2851 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
2852 (rs6000_expand_ternop_builtin): Likewise.
2853 * config/rs6000/rs6000-c.c
2854 (altivec_resolve_overloaded_builtin): Likewise.
2855 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
2856 (rs6000_expand_ternop_builtin): Likewise.
2857 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
2859 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2861 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
2862 when copying loop nest with only one inner loop.
2864 2017-10-10 Richard Biener <rguenther@suse.de>
2866 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
2867 blocks if SCEV is active.
2868 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
2870 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
2871 (scev_initialize): Assert we are not yet initialized.
2873 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2875 * tree-loop-distribution.c (generate_loops_for_partition): Remove
2876 inner loop's exit stmt by making it always exit the loop, otherwise
2877 we would generate an infinite empty loop.
2879 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2881 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
2882 renaming variables in new preheader if it's deleted.
2884 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2886 * tree-loop-distribution.c (struct partition): Remove unused field
2887 loops of the structure.
2888 (partition_alloc, partition_free): Ditto.
2889 (build_rdg_partition_for_vertex): Ditto.
2891 2017-10-09 Jeff Law <law@redhat.com>
2893 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
2894 return type to match prototype and documentation.
2896 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2898 * config/rs6000/rs6000.c (processor_costs): Move to ...
2899 * config/rs6000/rs6000.h: ... here.
2900 (rs6000_cost): Declare.
2902 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
2904 * except.c (setjmp_fn): New global variable.
2905 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
2906 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
2907 if DONT_USE_BUILTIN_SETJMP is defined.
2909 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2911 * target.def (insn_cost): New hook.
2912 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
2913 * doc/tm.texi: Regenerate.
2914 * rtlanal.c (insn_cost): Use the new hook.
2916 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2918 * combine.c (combine_validate_cost): Compute the new insn_cost,
2919 not just pattern_cost.
2920 (try_combine): Adjust comment.
2922 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2924 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
2926 * combine.c (uid_insn_cost): Adjust comment.
2927 (combine_validate_cost): Adjust comment. Use pattern_cost instead
2929 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
2930 * dse.c (find_shift_sequence): Ditto.
2931 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
2932 (bb_valid_for_noce_process_p): Use pattern_cost.
2933 * rtl.h (insn_rtx_cost): Delete.
2934 (pattern_cost): New prototype.
2935 (insn_cost): New prototype.
2936 * rtlanal.c (insn_rtx_cost): Rename to...
2937 (pattern_cost): ... this.
2940 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
2942 * config/i386/i386.md (*jcc_2): Remove insn pattern.
2943 (*jcc<mode>_0_r_i387): Ditto.
2944 (*jccxf_r_i387): Ditto.
2945 (*jcc<mode>_r_i387): Ditto.
2946 (*jccu<mode>_r_i387): Ditto.
2947 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
2948 (*jcc): Rename from *jcc_1.
2950 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2952 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
2953 deferred rescans after the lvx/stvx recombination pre-pass.
2955 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
2957 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
2958 memory operation instruction support.
2959 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
2960 (rs6000-ibm-aix[789]*): Likewise.
2961 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
2962 Document new functions.
2964 2017-10-09 Richard Biener <rguenther@suse.de>
2966 PR tree-optimization/82397
2967 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
2968 equality only for semantically equal trees.
2970 2017-10-09 Richard Biener <rguenther@suse.de>
2972 PR tree-optimization/82449
2973 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
2974 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
2975 allow constant addresses.
2976 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
2979 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2981 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
2984 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2987 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
2990 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2993 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
2995 2017-10-09 Jakub Jelinek <jakub@redhat.com>
2998 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
2999 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
3001 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
3003 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
3004 (WI_BINARY_PREDICATE_RESULT): Likewise.
3005 (wi::binary_traits::operator_result): New type.
3006 (wi::binary_traits::predicate_result): Likewise.
3007 (generic_wide_int::operator~, unary generic_wide_int::operator-)
3008 (generic_wide_int::operator==, generic_wide_int::operator!=)
3009 (generic_wide_int::operator&, generic_wide_int::and_not)
3010 (generic_wide_int::operator|, generic_wide_int::or_not)
3011 (generic_wide_int::operator^, generic_wide_int::operator+
3012 (binary generic_wide_int::operator-, generic_wide_int::operator*):
3014 (operator~, unary operator-, operator==, operator!=, operator&)
3015 (operator|, operator^, operator+, binary operator-, operator*): New
3017 * expr.c (get_inner_reference): Use wi::bit_and_not.
3018 * fold-const.c (fold_binary_loc): Likewise.
3019 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
3020 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
3021 (bit_value_binop): Likewise.
3022 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
3023 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
3024 (extract_range_from_binary_expr_1): Likewise.
3025 (masked_increment): Likewise.
3026 (simplify_bit_ops_using_ranges): Likewise.
3028 2017-10-09 Martin Jambor <mjambor@suse.cz>
3031 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
3032 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
3033 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
3035 (hsa_fixup_mov_insn_type): New function.
3036 (hsa_op_with_type::get_in_type): Use it.
3037 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
3038 immediates in an assert.
3039 (hsa_op_with_type::extend_int_to_32bit): New method.
3040 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
3041 types. Convert to dest type if necessary.
3042 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
3043 (reg_for_gimple_ssa): Pass false as min32int to
3044 hsa_type_for_scalar_tree_type.
3045 (gen_hsa_addr): Fixup type when creating addresable temporary.
3046 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
3047 (gen_hsa_unary_operation): Extend operands and convert to dest type if
3048 necessary. Call hsa_fixup_mov_insn_type.
3049 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
3050 extend operands and convert to dest type if necessary.
3051 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
3052 to dest type if necessary.
3053 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
3054 if conversion nt necessary and size matches.
3055 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
3056 to dest type if necessary.
3057 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
3058 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
3060 (gen_hsa_clrsb): Likewise.
3061 (gen_hsa_ffs): Likewise.
3062 (gen_hsa_divmod): Extend operands and convert to dest type if
3064 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
3066 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3068 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
3069 Remove empty default arguments. Use a brace block as output
3071 (conditional return): Ditto.
3073 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
3074 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
3076 (group_ending_nop): Ditto.
3077 (doloop_end): Ditto.
3078 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
3079 (splitters for those): Ditto.
3081 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3083 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
3084 a conditional jump (and the compare for it) so that pc_rtx is the
3086 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
3087 for the deleted and renamed ctr<mode>_internal[234] patterns.
3088 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
3089 Delete second conditional return pattern.
3090 (ctr<mode>_internal2): Delete this second bdnz pattern.
3091 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
3092 (ctr<mode>_internal4): Delete this second bdz pattern.
3094 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3096 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
3097 (always_initialized_rtx_for_ssa_name_p): New predicate.
3098 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
3099 (finish_out_of_ssa): Free new field of SA.
3100 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
3101 * tree-ssa-coalesce.c: Include tree-ssa.h.
3102 (get_parm_default_def_partitions): Remove extern keyword.
3103 (get_undefined_value_partitions): New function.
3104 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
3105 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
3107 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
3109 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3111 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
3113 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3115 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
3116 for targets that preffer 128bit.
3118 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3120 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
3122 2017-10-08 Olivier Hainque <hainque@adacore.com>
3124 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
3125 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
3127 (thumb_set_return_address): Likewise.
3129 2017-10-08 Olivier Hainque <hainque@adacore.com>
3131 * common/config/arm/arm-common.c (arm_except_unwind_info):
3132 Handle DWARF2_UNWIND_INFO.
3134 2017-10-07 Michael Collison <michael.collison@arm.com>
3136 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
3139 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
3141 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
3142 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
3143 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
3144 defined, force the creation of a new block for a dispatch label.
3146 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
3148 * invoke.texi (Wsuggest-attribute=cold): Document.
3149 * common.opt (Wsuggest-attribute=cold): New
3150 * ipa-pure-const.c (warn_function_cold): New function.
3151 * predict.c (compute_function_frequency): Use it.
3152 * predict.h (warn_function_cold): Declare.
3154 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
3156 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
3159 2017-10-06 Martin Liska <mliska@suse.cz>
3161 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
3162 keyword for member functions.
3163 (struct sanopt_tree_couple): New struct.
3164 (struct sanopt_tree_couple_hash): New function.
3165 (struct sanopt_ctx): Add new hash_map.
3166 (has_dominating_ubsan_ptr_check): New function.
3167 (record_ubsan_ptr_check_stmt): Likewise.
3168 (maybe_optimize_ubsan_ptr_ifn): Likewise.
3169 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
3170 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
3172 2017-10-06 Sudakshina Das <sudi.das@arm.com>
3175 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
3176 aarch64_simd_valid_immediate on CONST_VECTORs.
3177 (aarch64_reg_or_bic_imm): Likewise.
3179 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
3181 PR rtl-optimization/82396
3182 * haifa-sched.c (ready_sort_real): Disable qsort checking.
3184 2017-10-06 Sebastian Pop <sebpop@gmail.com>
3186 * graphite-dependences.c (scop_get_reads): Move code to...
3187 (scop_get_must_writes): Move code to...
3188 (scop_get_may_writes): Move code to...
3189 (scop_get_reads_and_writes): ... here.
3190 (scop_get_dependences): Call scop_get_reads_and_writes.
3192 2017-10-06 Jakub Jelinek <jakub@redhat.com>
3194 PR tree-optimization/82434
3195 * fold-const.h (can_native_encode_type_p,
3196 can_native_encode_string_p): Remove.
3197 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
3198 don't encode anything, just return what would be otherwise returned.
3199 (native_encode_fixed, native_encode_complex, native_encode_vector):
3201 (native_encode_string): Likewise. Inline by hand
3202 can_native_encode_string_p.
3203 (can_native_encode_type_p): Remove.
3204 (can_native_encode_string_p): Remove.
3205 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
3206 STRING_CSTs using can_native_encode_string_p, test all
3207 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
3208 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
3209 argument from native_encode_expr.
3210 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
3211 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
3214 2017-10-06 Richard Biener <rguenther@suse.de>
3216 PR tree-optimization/82397
3217 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
3218 operand_equal_p but rely on data_ref_compare_tree for detecting
3220 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
3221 to match up with dr_group_sort_cmp.
3223 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3226 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
3228 * config/s390/s390-builtin-types.def: Regenerate.
3230 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3233 * config/s390/s390-builtin-types.def: Regenerate.
3234 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
3235 Change flag from B_VXE to B_VX.
3236 (s390_vec_min_dbl): Remove B_VXE flag.
3238 2017-10-06 Richard Biener <rguenther@suse.de>
3240 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
3241 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
3242 translate_isl_ast_to_gimple::is_valid_rename,
3243 translate_isl_ast_to_gimple::get_rename,
3244 translate_isl_ast_to_gimple::get_def_bb_for_const,
3245 translate_isl_ast_to_gimple::get_new_name,
3246 translate_isl_ast_to_gimple::collect_all_ssa_names,
3247 translate_isl_ast_to_gimple::copy_loop_phi_args,
3248 translate_isl_ast_to_gimple::collect_all_ssa_names,
3249 translate_isl_ast_to_gimple::copy_loop_phi_args,
3250 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
3251 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
3252 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
3253 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
3254 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
3255 translate_isl_ast_to_gimple::copy_cond_phi_args,
3256 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
3257 translate_isl_ast_to_gimple::edge_for_new_close_phis,
3258 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
3259 translate_isl_ast_to_gimple::rename_uses,
3260 translate_isl_ast_to_gimple::rename_all_uses): Remove.
3261 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
3262 (set_rename_for_each_def): Likewise.
3263 (graphite_copy_stmts_from_block): Handle debug stmt resetting
3264 here. Handle rewriting SCEV analyzable uses here.
3265 (copy_bb_and_scalar_dependences): Generate code for PHI
3267 (graphite_regenerate_ast_isl): Adjust.
3268 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
3269 (add_write, add_read): New functions.
3270 (build_cross_bb_scalars_def): Use it and simplify.
3271 (build_cross_bb_scalars_use): Likewise.
3272 (graphite_find_cross_bb_scalar_vars): Inline into...
3273 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
3274 simulating out-of-SSA. Compute liveout and add dependencies.
3275 (build_scops): Force an empty entry block.
3276 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
3278 (sese_build_liveouts): Declare.
3279 (sese_trivially_empty_bb_p): Likewise.
3280 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
3281 compute liveout and debug_liveout.
3282 (sese_bad_liveouts_use): Remove.
3283 (sese_reset_debug_liveouts_bb): Likewise.
3284 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
3285 (sese_build_liveouts): Build liveout and debug_liveout and store
3287 (new_sese_info): Adjust.
3288 (free_sese_info): Likewise.
3289 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
3290 do not build liveout here.
3291 (move_sese_in_condition): Adjust region entry.
3292 (scev_analyzable_p): Match up with chrec_apply requirements.
3293 (sese_trivially_empty_bb_p): New.
3294 * tree-into-ssa.c (get_reaching_def): Properly support generating
3295 default-defs for incremental rewrite of anonymous names.
3297 2017-10-06 Richard Biener <rguenther@suse.de>
3299 * graphite-sese-to-poly.c (extract_affine): For casts increasing
3300 precision do not perform modulo reduction.
3302 2017-10-06 Richard Biener <rguenther@suse.de>
3304 PR tree-optimization/82436
3305 * tree-vect-slp.c (vect_supported_load_permutation_p): More
3306 conservatively choose the vectorization factor when checking
3307 whether we can perform the required load permutation.
3308 (vect_transform_slp_perm_load): Assert when we may not fail.
3310 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
3312 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
3313 message for incompatible -msdata=* and -mcall-* options.
3315 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3317 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
3318 rate for post-reload scheduling.
3320 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3322 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
3324 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3326 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
3327 to improve monte carlo in scimark.
3329 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3331 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
3332 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
3333 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
3334 pentium4_cost, nocona_cost): Set reassociation width to 1.
3335 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
3336 width to 2 for fp operations and 1 otherwise.
3337 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
3339 (atom_cost): Set reassociation width to 2.
3340 (slm_cost, generic_cost): Set fp reassociation width
3341 to 2 and 1 otherwise.
3342 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
3343 (core_cost): Set fp reassociation width to 4 and vector to 2.
3344 (ix86_reassociation_width): Rewrite using cost table; special case
3345 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
3346 and TARGET_AVX128_OPTIMAL.
3347 * config/i386/i386.h (processor_costs): Add
3348 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
3349 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
3350 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
3351 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
3352 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
3353 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
3355 2017-10-05 Nathan Sidwell <nathan@acm.org>
3357 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
3359 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3361 * config/arm/arm.c (arm_test_fpu_data): New.
3362 (arm_run_selftests): Call arm_test_fpu_data.
3364 2017-10-04 Nathan Sidwell <nathan@acm.org>
3366 * toplev.c (toplev::main): Remove excess parens on pretty_printer
3368 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
3370 2017-10-04 Sudakshina Das <sudi.das@arm.com>
3372 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
3373 check type for aarch64_simd_valid_immediate.
3374 (aarch64_output_simd_mov_immediate): Update prototype.
3375 (aarch64_simd_valid_immediate): Update prototype.
3376 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
3377 support for ORR-immediate.
3378 (and<mode>3): modified pattern to add support for BIC-immediate.
3379 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
3380 now checks for valid immediate for BIC and ORR based on new enum
3382 (aarch64_output_simd_mov_immediate): Function now used to output
3383 BIC/ORR imm as well based on new enum argument.
3384 * config/aarch64/constraints.md (Do): New vector immediate constraint.
3386 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
3387 (aarch64_reg_or_bic_imm): Likewise.
3389 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3391 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
3392 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
3394 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3398 PR rtl-optimization/82396
3399 * haifa-sched.c (autopref_multipass_init): Simplify
3401 (autopref_rank_data): Simplify sort order.
3402 * sched-int.h (autopref_multipass_data_): Remove
3403 multi_mem_insn_p, min_offset and max_offset.
3405 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3407 * doc/sourcebuild.texi: Document vect_peeling_profitable.
3409 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3411 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
3414 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3416 * doc/sourcebuild.texi: Document vect_long_mult.
3418 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
3420 PR tree-optimization/82413
3421 * fold-const.c (build_range_check): Use widest_int when comparing
3422 the maximum ETYPE value with HIGH.
3424 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3426 PR rtl-optimization/82396
3427 * haifa-sched.c (autopref_multipass_init): Simplify
3429 (autopref_rank_data): Simplify sort order.
3430 * sched-int.h (autopref_multipass_data_): Remove
3431 multi_mem_insn_p, min_offset and max_offset.
3433 2017-10-04 Jakub Jelinek <jakub@redhat.com>
3435 PR tree-optimization/82381
3436 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
3437 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
3440 PR tree-optimization/82374
3441 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
3442 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
3443 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
3444 current_function_decl to the new decl.
3446 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
3448 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
3449 helper macro for IEEE float128 hardware built-in functions.
3450 (SQRTF128_ODD): Add built-in functions with the round-to-odd
3452 (TRUNCF128_ODD): Likewise.
3453 (ADDF128_ODD): Likewise.
3454 (SUBF128_ODD): Likewise.
3455 (MULF128_ODD): Likewise.
3456 (DIVF128_ODD): Likewise.
3457 (FMAF128_ODD): Likewise.
3458 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
3459 UNSPEC_TRUNC_ROUND_TO_ODD.
3460 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
3461 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
3462 floating point round to odd instructions.
3463 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
3464 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
3465 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
3466 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
3467 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
3468 (trunc<mode>sf2_hw): Change the truncate with round to odd
3469 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
3470 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
3471 to odd hardware instructions.
3472 (sub<mode>3_odd): Likewise.
3473 (mul<mode>3_odd): Likewise.
3474 (div<mode>3_odd): Likewise.
3475 (sqrt<mode>2_odd): Likewise.
3476 (fma<mode>4_odd): Likewise.
3477 (fms<mode>4_odd): Likewise.
3478 (nfma<mode>4_odd): Likewise.
3479 (nfms<mode>4_odd): Likewise.
3480 (trunc<mode>df2_odd): Change the truncate with round to odd
3481 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
3483 * doc/extend.texi (PowerPC built-in functions): Update documentation
3484 for existing IEEE float128-bit built-in functions. Add built-in
3485 functions that generate the IEEE 128-bit floating point round to
3488 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
3490 PR rtl-optimization/77729
3491 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
3492 to (X&(C1&~C2))|C2 transformations.
3494 2017-10-03 Martin Jambor <mjambor@suse.cz>
3496 PR tree-optimization/82363
3497 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
3498 mismatch, mark lacc written regardless of racc.
3500 2017-10-03 Jakub Jelinek <jakub@redhat.com>
3502 PR tree-optimization/82381
3503 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
3504 stmt_to_insert nor wheather SSA_NAMEs are default defs.
3505 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
3506 fallthrough into reassoc_stmt_dominates_stmt_p.
3509 * combine.c (combine_instructions): Don't combine in unreachable
3512 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
3515 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
3516 function to not use the have_cpu variable. Do not set cpu_index,
3517 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
3519 (rs6000_valid_attribute_p): Remove duplicate initializations of
3520 old_optimize and func_optimize.
3521 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
3522 (rs6000_activate_target_options): Make global.
3523 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
3526 2017-10-02 Jakub Jelinek <jakub@redhat.com>
3528 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
3529 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
3530 Set *poffset to 0 and *psize and *pmax_size to -1 if
3531 *poffset + *psize overflows in HOST_WIDE_INT.
3533 PR tree-optimization/82387
3534 PR tree-optimization/82388
3535 PR tree-optimization/82389
3536 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
3537 instead of live_bytes non-NULL.
3539 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
3542 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
3545 2017-10-02 Richard Biener <rguenther@suse.de>
3547 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
3548 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
3549 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
3551 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3553 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
3554 requested precision matches the type's.
3555 * calls.c (alloc_max_size): Calculate the new candidate size as
3556 a widest_int and use wi::to_widest when comparing it with the
3557 current candidate size.
3558 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
3559 zero rather than integer_zero_node.
3560 * match.pd: Check for a no-op conversion before using wi::add
3561 rather than after. Use tree_to_uhwi when summing small shift
3562 counts into an unsigned int.
3564 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3565 Alan Hayward <alan.hayward@arm.com>
3566 David Sherwood <david.sherwood@arm.com>
3569 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
3570 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
3571 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
3572 POINTER_AND_FP_REGS.
3574 2017-10-02 Richard Biener <rguenther@suse.de>
3576 PR tree-optimization/82355
3577 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3578 a mapping for the enclosing loop but avoid generating one for
3580 (copy_bb_and_scalar_dependences): Remove premature codegen
3581 error on PHIs in blocks duplicated into multiple places.
3582 * graphite-scop-detection.c
3583 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3584 in the region use it as loop and nest to analyze the DR in.
3585 (try_generate_gimple_bb): Likewise.
3586 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3587 (add_loop_constraints): For blocks in a loop not in the region
3588 create a dimension with a single iteration.
3589 * sese.h (gbb_loop_at_index): Remove assert.
3591 2017-10-01 Kevin Buettner <kevinb@redhat.com>
3593 * omp-expand.c (adjust_context_scope): New function.
3594 (expand_parallel_call): Call adjust_context_scope.
3596 2017-10-01 Jeff Law <law@redhat.com>
3598 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
3599 dom_opt_dom_walker class with direct access to private members.
3600 Add comments. Call test_for_singularity.
3601 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
3602 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
3603 m_dummy_cond anymore.
3604 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
3606 (pass_dominator:execute): Build the dummy_cond here and pass it
3607 to the dom_opt_dom_walker ctor.
3608 (test_for_singularity): New function.
3610 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
3611 Maya Rashish <coypu@sdf.org>
3613 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
3614 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
3615 (alpha*-*-netbsd*) Use nbsd_tm_file.
3616 (arm*-*-netbsdelf*) Likewise.
3617 (i[34567]86-*-netbsdelf*) Likewise.
3618 (x86_64-*-netbsd*) Likewise.
3619 (mips*-*-netbsd*) Likewise.
3620 (powerpc-*-netbsd*) Likewise.
3621 (sh*-*-netbsd*) Likewise.
3622 (sparc-*-netbsdelf*) Likewise.
3623 (sparc64-*-netbsd*) Likewise.
3624 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
3626 (vax-*-netbsdelf*) Likewise.
3627 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
3628 (UINT_FAST8_TYPE) Likewise.
3629 (INT_FAST16_TYPE) Check CHAR_FAST16.
3630 (UINT_FAST16_TYPE) Likewise.
3632 2017-09-30 Jakub Jelinek <jakub@redhat.com>
3635 * config/i386/i386.md
3636 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
3637 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
3638 *divmodsi4_zext_2): New define_insn_and_split.
3639 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
3640 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
3641 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
3642 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
3643 New define_insn_and_split.
3644 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
3645 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
3646 operands[1] having DImode when mode is SImode.
3648 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
3649 always SImode for DIV and MOD in REG_EQUAL notes.
3651 2017-09-29 Yury Gribov <tetra2005@gmail.com>
3654 * match.pd: Fix handling of NaNs in pattern.
3656 2017-09-29 Jeff Law <law@redhat.com>
3658 * sbitmap.c (bitmap_bit_in_range_p): New function.
3659 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
3660 * tree-ssa-dse.c (live_bytes_read): New function.
3661 (dse_classify_store): Ignore reads of dead bytes.
3663 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
3664 typos and whitespace errors.
3665 * config/i386/predicates.md (address_no_seg_operand): Likewise.
3666 * config/s390/s390.c (s390_emit_prologue): Likewise.
3668 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3671 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
3672 with a symbol for LRA.
3674 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3676 PR rtl-optimization/82338
3677 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
3679 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
3681 * genmodes.c (calc_wider_mode): Suppress qsort macro.
3682 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
3683 (qsort_chk): Declare.
3684 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
3685 (qsort_chk): New function.
3687 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3689 PR tree-optimization/82337
3690 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
3691 phi definition if the PHI result appears in an abnormal PHI.
3692 (find_basis_for_base_expr): Don't record a basis if the LHS of the
3693 basis appears in an abnormal PHI.
3695 2017-09-29 Richard Biener <rguenther@suse.de>
3697 * graphite-isl-ast-to-gimple.c
3698 (translate_isl_ast_to_gimple::set_codegen_error): New function.
3699 (binary_op_to_tree): Use it.
3700 (get_rename_from_scev): Likewise.
3701 (copy_loop_phi_nodes): Likewise.
3702 (copy_bb_and_scalar_dependences): Likewise.
3703 (translate_pending_phi_nodes): Likewise.
3705 2017-09-29 Jakub Jelinek <jakub@redhat.com>
3708 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
3709 for movabsq $(i32 << shift), r64.
3711 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
3713 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
3714 index when encoding %esp as %rsp to avoid 0x67 prefix.
3716 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
3718 * config/i386/i386.md (*movsf_internal, *movdf_internal):
3719 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
3721 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
3723 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
3724 Extensions with more than 16 double VFP registers.
3725 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
3726 to_clear_mask and all code related to it. Replace the remaining
3727 entry by a sbitmap and adapt code accordingly.
3729 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
3731 * brig-builtins.def: Change pure attributes to const.
3733 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3735 * config.gcc (default_gnu_indirect_function): Default to yes for
3736 sparc*-*-linux* with glibc.
3738 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3740 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
3741 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
3742 when creating .init_array and .fini_array sections with priority
3745 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
3748 * config/aarch64/aarch64.c
3749 (aarch64_builtin_support_vector_misalignment): Always return false
3750 when misalignment is unknown.
3752 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
3754 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
3755 this function to return false if the definition used by the swap
3756 instruction is artificial, or if the memory address from which the
3757 constant value is loaded is not represented by a base address held
3758 in a register or if the base address register is a frame or stack
3759 pointer. Additionally, return false if the base address of the
3760 loaded constant is a SYMBOL_REF but is not considered to be a
3762 (replace_swapped_load_constant): New function.
3763 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
3764 loaded constant vector with a load of a swapped constant vector.
3766 2017-09-27 Carl Love <cel@us.ibm.com>
3768 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
3769 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
3770 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
3773 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
3775 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
3776 first, always call autopref_rank_data otherwise.
3778 2017-09-27 Richard Biener <rguenther@suse.de>
3780 * graphite-scop-detection.c (find_scop_parameters): Move
3781 loop bound handling ...
3782 (gather_bbs::before_dom_children): ... here, avoiding the need
3783 to build scop_info->loop_nest.
3784 (record_loop_in_sese): Remove.
3785 * sese.h (sese_info_t::loop_nest): Remove.
3786 * sese.c (new_sese_info): Do not allocate loop_nest.
3787 (free_sese_info): Do not free loop_nest.
3789 2017-09-27 Jakub Jelinek <jakub@redhat.com>
3792 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
3793 lhs from calls if the lhs has addressable type.
3795 2017-09-27 Richard Biener <rguenther@suse.de>
3797 * graphite.h (scop::max_alias_set): New member.
3798 * graphite-scop-detection.c: Remove references to non-existing
3799 --param in comments.
3800 (build_alias_sets): Record the maximum alias set used for drs.
3801 (build_scops): Support zero as unlimited for
3802 --param graphite-max-arrays-per-scop.
3803 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
3805 (build_poly_sr_1): ... here. Compute alias set based on the
3806 maximum alias set used for drs rather than
3807 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
3809 2017-09-27 Richard Biener <rguenther@suse.de>
3811 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
3812 --param loop-block-tile-size=0 to disable tiling.
3814 2017-09-27 Richard Biener <rguenther@suse.de>
3816 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
3817 (graphite-max-nb-scop-params): Document special value zero.
3818 * domwalk.h (dom_walker::STOP): New symbolical constant.
3819 (dom_walker::dom_walker): Add optional parameter for bb to
3821 (dom_walker::~dom_walker): Declare.
3822 (dom_walker::before_dom_children): Document STOP return value.
3823 (dom_walker::m_user_bb_to_rpo): New member.
3824 (dom_walker::m_bb_to_rpo): Likewise.
3825 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
3826 mapping here if not provided by the user.
3827 (dom_walker::~dom_walker): Free bb to RPO mapping if not
3828 provided by the user.
3829 (dom_walker::STOP): Define.
3830 (dom_walker::walk): Do not compute bb to RPO mapping here.
3831 Support STOP return value from before_dom_children to stop
3833 * graphite-optimize-isl.c (optimize_isl): If the schedule
3834 is the same still generate code if -fgraphite-identity
3835 or -floop-parallelize-all are given.
3836 * graphite-scop-detection.c: Include cfganal.h.
3837 (gather_bbs::gather_bbs): Get and pass through bb to RPO
3839 (gather_bbs::before_dom_children): Return STOP for BBs
3841 (build_scops): Compute bb to RPO mapping and pass it to
3842 the domwalk. Treat --param graphite-max-nb-scop-params=0
3843 as not limiting the number of params.
3844 * graphite.c (graphite_initialize): Remove limit on the
3845 number of basic-blocks in a function.
3846 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
3847 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
3848 default value of 10.
3850 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3852 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
3853 Adjust code to eliminate needing to do the shift right 32-bits
3854 operation after XSCVDPSPN.
3856 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3858 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
3859 ((X / Y) != 0 -> X >= Y): Likewise.
3861 2017-09-26 Carl Love <cel@us.ibm.com>
3863 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
3864 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
3865 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
3866 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
3867 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
3868 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
3869 definitions and overloading.
3870 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
3871 statement for P9V_BUILTIN_XST_LEN_R.
3872 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
3873 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
3874 define_expand and define_insn for the instructions and builtins.
3875 * doc/extend.texi: Update the built-in documentation file for the new
3877 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
3878 define_insn for the instructions
3880 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
3883 * gcc/config/netbsd-protos.h: New file.
3884 * gcc/config/netbsd.c: New file.
3885 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
3886 * gcc/config/t-netbsd: New file.
3887 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
3888 (tmake_file) Add t-netbsd.
3889 (extra_objs) Add netbsd.o.
3891 2017-09-26 Janus Weil <janus@gcc.gnu.org>
3895 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
3897 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3899 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
3900 sign extension from a vector register to a GPR by doing a 32-bit
3901 direct move and then an EXTSW.
3902 (extendsi<mode>2 splitter): Likewise.
3903 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
3904 right or vector extract after doing XSCVDPSPN. Use
3905 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
3907 (movdi_from_sf_zero_ext): Likewise.
3908 (reload_gpr_from_vsxsf): Likewise.
3909 (p8_mfvsrd_4_disf): Delete, no longer used.
3910 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
3911 then needing to move the SFmode to a GPR to use the XSCVDPSP
3912 instruction instead of FRSP and XSCVDPSPN.
3913 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
3914 it is adjacent to the other XSCVSPDP insns.
3915 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
3916 SFmode to be in traditional Altivec registers.
3917 (vsx_xscvdpspn): Eliminate useless alternative constraint.
3918 (vsx_xscvspdpn): Likewise.
3919 (vsx_xscvspdpn_scalar): Likewise.
3921 2017-09-26 Martin Jambor <mjambor@suse.cz>
3923 * tree-sra.c (compare_access_positions): Put integral types first,
3924 stabilize sorting of integral types, remove conditions putting
3925 non-full-precision integers last.
3926 (sort_and_splice_var_accesses): Disable scalarization if a
3927 non-integert would be represented by a non-full-precision integer.
3929 2017-09-26 Joseph Myers <joseph@codesourcery.com>
3931 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
3932 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
3933 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
3934 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
3935 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
3936 conditionals inside the function instead of around it. Call
3937 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
3938 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
3940 2017-09-26 Richard Biener <rguenther@suse.de>
3942 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
3944 (scop_detection::build_scop_breadth): ... this. Removed.
3945 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
3946 (scop_detection::harmful_stmt_in_bb): Likewise.
3947 (scop_detection::graphite_can_represent_stmt): Likewise.
3948 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
3949 (scop_detection::can_represent_loop): Remove recursion, fold in ...
3950 (scop_detection::can_represent_loop_1): ... this. Removed.
3951 (scop_detection::harmful_loop_in_region): Simplify after inlining
3952 the above and remove more quadraticness.
3953 (build_scops): Adjust.
3954 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
3957 2017-09-26 Jakub Jelinek <jakub@redhat.com>
3960 * config/i386/i386.c (ix86_print_operand_address_as): Only test
3961 REGNO (base) == SP_REG if base is a REG.
3964 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
3965 if it is different SSA_NAME.
3966 (optimize_range_tests_cmp_bitwise): New function.
3967 (optimize_range_tests): Call it.
3969 2017-09-26 Richard Biener <rguenther@suse.de>
3971 PR tree-optimization/82321
3972 * graphite.c (canonicalize_loop_closed_ssa): Properly check
3973 for the def being inside the loop.
3975 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3977 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
3979 * config/s390/s390-builtins.def: Fix constraint on op4.
3981 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3983 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
3984 independent expanders.
3985 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
3986 ("vec_ordered", "vec_unordered"): New expanders.
3988 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3990 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
3993 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3995 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
3996 vec_unpacks_lo_v16qi.
3997 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
3999 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4001 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
4002 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
4003 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
4005 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4007 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
4009 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
4011 ("*vec_slb<mode>"): New insn pattern.
4012 ("vec_shr_<mode>"): New expander.
4013 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
4014 and force the shift count operand to V16QImode.
4015 ("vec_srb<mode>"): Set shift count mode to V16QI.
4017 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4019 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
4020 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
4021 ("vec_widen_smult_hi_<mode>"): New expander definitions.
4023 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
4026 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
4028 2017-09-26 Richard Biener <rguenther@suse.de>
4030 PR tree-optimization/82320
4031 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
4034 2017-09-25 Jeff Law <law@redhat.com>
4036 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
4037 prototype for new argument.
4038 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
4039 mostly extracted from rs6000_emit_allocate_stack.
4040 (rs6000_emit_probe_stack_range_stack_clash): New function.
4041 (rs6000_emit_allocate_stack): Call
4042 rs6000_emit_probe_stack_range_stack_clash as needed.
4043 (rs6000_emit_probe_stack_range): Add additional argument
4044 to call to gen_probe_stack_range{si,di}.
4045 (output_probe_stack_range): New.
4046 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
4047 (output_probe_stack_range_stack_clash): New.
4048 (rs6000_emit_prologue): Emit notes into dump file as requested.
4049 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
4050 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
4051 Add additional operand and pass it to output_probe_stack_range.
4053 2017-09-25 Bin Cheng <bin.cheng@arm.com>
4055 PR tree-optimization/82163
4056 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
4057 (checking_verify_loop_closed_ssa): New parameter.
4058 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
4059 (check_loop_closed_ssa_stmt): Delete.
4060 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
4061 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
4062 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
4065 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
4067 * brig-builtins.def: Treat HSAIL barrier builtins as
4068 setjmp/longjump style functions.
4070 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
4072 * target.def (constant_alignment): New hook.
4073 * defaults.h (CONSTANT_ALIGNMENT): Delete.
4074 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
4075 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
4076 * doc/tm.texi: Regenerate.
4077 * targhooks.h (default_constant_alignment): Declare.
4078 (constant_alignment_word_strings): Likewise.
4079 * targhooks.c (default_constant_alignment): New function.
4080 (constant_alignment_word_strings): Likewise.
4081 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
4082 instead of CONSTANT_ALIGNMENT.
4083 * varasm.c (align_variable, get_variable_align, build_constant_desc)
4084 (force_const_mem): Likewise.
4085 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
4086 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
4087 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
4088 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4089 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
4091 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
4092 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4093 constant_alignment_word_strings.
4094 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
4095 (CONSTANT_ALIGNMENT): Likewise.
4096 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4097 (arm_constant_alignment): New function.
4098 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
4099 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4100 constant_alignment_word_strings.
4101 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
4102 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4103 constant_alignment_word_strings.
4104 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
4105 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4106 (cris_constant_alignment): New function.
4107 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
4108 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4109 (epiphany_constant_alignment): New function.
4110 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
4111 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4112 constant_alignment_word_strings.
4113 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
4114 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4115 constant_alignment_word_strings.
4116 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
4117 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4118 constant_alignment_word_strings.
4119 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
4120 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
4121 * config/i386/i386.c (ix86_constant_alignment): Make static.
4122 Use the same interface as the target hook.
4123 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4124 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
4125 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4126 constant_alignment_word_strings.
4127 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
4128 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
4129 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4130 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
4131 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4132 constant_alignment_word_strings.
4133 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
4134 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4135 constant_alignment_word_strings.
4136 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
4137 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4138 constant_alignment_word_strings.
4139 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
4140 * config/microblaze/microblaze.c (microblaze_constant_alignment):
4142 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4143 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
4144 * config/mips/mips.c (mips_constant_alignment): New function.
4145 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4146 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
4147 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
4148 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4149 (mmix_constant_alignment): Make static. Use the same interface
4151 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
4152 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4153 constant_alignment_word_strings.
4154 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
4155 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4156 constant_alignment_word_strings.
4157 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
4158 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4159 constant_alignment_word_strings.
4160 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
4161 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4162 (rs6000_constant_alignment): New function.
4163 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
4164 * config/riscv/riscv.c (riscv_constant_alignment): New function.
4165 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4166 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
4167 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4168 (rs6000_constant_alignment): New function.
4169 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
4170 * config/s390/s390.c (s390_constant_alignment): New function.
4171 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4172 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
4173 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4174 constant_alignment_word_strings.
4175 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
4176 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4177 (sparc_constant_alignment): New function.
4178 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
4179 * config/spu/spu.c (spu_constant_alignment): New function.
4180 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4181 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
4182 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4183 constant_alignment_word_strings.
4184 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
4185 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4186 constant_alignment_word_strings.
4187 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
4188 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4189 constant_alignment_word_strings.
4190 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
4191 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4192 (visium_constant_alignment): New function.
4193 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
4194 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4195 (xtensa_constant_alignment): New function.
4196 * system.h (CONSTANT_ALIGNMENT): Poison.
4198 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
4200 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4201 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
4202 (rs6000_builtin_valid_without_lhs): New helper function.
4203 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4204 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
4206 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
4208 * target.h (vec_perm_indices): Use unsigned short rather than
4210 (auto_vec_perm_indices): Likewise.
4211 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
4212 Use unsigned int rather than unsigned char.
4213 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
4215 2017-09-25 Richard Biener <rguenther@suse.de>
4217 * cfgloop.h (sort_sibling_loops): Declare.
4218 * cfgloop.c (sort_sibling_loops_cmp): New helper.
4219 (sort_sibling_loops): New function sorting the sibling loop list
4221 * graphite.c (graphite_transform_loops): Sort sibling loops.
4223 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
4225 * target.def (vec_perm_const_ok): Change sel parameter to
4227 * optabs-query.c (can_vec_perm_p): Update accordingly.
4228 * doc/tm.texi: Regenerate.
4229 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
4230 auto_vec_perm_indices and remove separate nelt field.
4231 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
4232 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
4233 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
4234 (aarch64_expand_vec_perm_const): Update accordingly.
4235 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
4236 to vec_perm_indices.
4237 * config/arm/arm.c (expand_vec_perm_d): Change perm to
4238 auto_vec_perm_indices and remove separate nelt field.
4239 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
4240 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
4241 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
4243 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
4244 to vec_perm_indices.
4245 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
4246 sel to vec_perm_indices.
4247 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
4248 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
4249 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
4251 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
4254 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
4257 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
4258 on the FUNCTION_DECL function context if it has a DIE that is a
4261 2017-09-25 Richard Biener <rguenther@suse.de>
4263 PR tree-optimization/82285
4264 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
4267 2017-09-25 Tom de Vries <tom@codesourcery.com>
4271 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
4274 2017-09-25 Richard Biener <rguenther@suse.de>
4276 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
4277 ISL errors other than isl_error_quota happen. Dump if the
4278 schedule is the same.
4279 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
4280 errors instead of aborting inside ISL.
4282 2017-09-25 Iain Sandoe <iain@codesourcery.com>
4285 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
4286 of libgcc_eh for m64.
4287 * config/i386/darwin64.h: Likewise.
4289 2017-09-25 Richard Biener <rguenther@suse.de>
4292 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
4293 attribute for incomplete types nor twice for complete ones.
4295 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
4298 * config/i386/i386.c (ix86_print_operand_address_as): Encode
4299 %esp as %rsp to avoid 0x67 prefix if there is no index or base
4302 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
4305 * config/i386/i386.opt (mprefer-avx256): Use
4306 ix86_target_flags variable.
4307 * config/i386/i386.c (ix86_target_string): Move
4308 -mprefer-avx256 to flag2_opts.
4310 2017-09-22 Jakub Jelinek <jakub@redhat.com>
4313 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
4314 and x != -1 | y != -1 into (x & y) != -1.
4316 2017-09-22 Steve Ellcey <sellcey@cavium.com>
4318 * config.gcc: Add new case statement to set
4319 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
4320 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
4321 s390x-*-linux* case statements. Added aarch64 to the list of
4322 supported architectures.
4324 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4326 PR tree-optimization/82289
4327 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
4328 STMT_VINFO_RELEVANT_P.
4330 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4331 Alan Hayward <alan.hayward@arm.com>
4332 David Sherwood <david.sherwood@arm.com>
4334 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
4335 for VR_RANGE only; don't allow VR_ANTI_RANGE.
4336 (extract_range_from_binary_expr_1): Don't call
4337 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
4339 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4340 Alan Hayward <alan.hayward@arm.com>
4341 David Sherwood <david.sherwood@arm.com>
4343 * target.def (preferred_vector_alignment): New hook.
4344 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
4346 * doc/tm.texi: Regenerate.
4347 * targhooks.h (default_preferred_vector_alignment): Declare.
4348 * targhooks.c (default_preferred_vector_alignment): New function.
4349 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
4351 (DR_TARGET_ALIGNMENT): New macro.
4352 (aligned_access_p): Update commentary.
4353 (vect_known_alignment_in_bytes): New function.
4354 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
4356 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
4357 Calculate the misalignment based on the target alignment rather than
4359 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
4360 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
4361 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
4362 the target alignment, rather than masking the element misalignment
4363 with the number of elements in a vector. Also use the target
4364 alignment when calculating the maximum number of peels.
4365 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
4366 instead of TYPE_ALIGN_UNIT.
4367 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
4368 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
4369 (vect_create_addr_base_for_vector_ref): Update call accordingly.
4370 (vect_create_data_ref_ptr): Likewise.
4371 (vect_setup_realignment): Realign by ANDing with
4372 -DR_TARGET_MISALIGNMENT.
4373 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
4374 the number of peels based on DR_TARGET_ALIGNMENT.
4375 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
4376 with the guaranteed alignment boundary when deciding whether
4378 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
4379 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
4380 (ensure_base_align): Remove stmt_info parameter. Get the
4381 target base alignment from DR_TARGET_ALIGNMENT.
4382 (vectorizable_store): Update call accordingly. Interpret
4383 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
4385 (vectorizable_load): Likewise.
4387 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4388 Alan Hayward <alan.hayward@arm.com>
4389 David Sherwood <david.sherwood@arm.com>
4391 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
4392 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
4393 (vect_enhance_data_refs_alignment): Likewise.
4395 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
4397 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4398 error. Only quit immediately if parsing is complete.
4399 (BEGIN): Initialize fatal_err and parse_done.
4400 (begin fpu, end fpu): Check number of arguments.
4401 (begin arch, end arch): Likewise.
4402 (begin cpu, end cpu): Likewise.
4403 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4404 (optalias): Likewise.
4406 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
4408 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4409 * config/arm/arm-isa.h: Delete. Move definitions to ...
4410 * arm-cpus.in: ... here. Use new feature and fgroup values.
4411 * config/arm/arm.c (arm_option_override): Use lower case for feature
4413 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4414 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4415 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4417 (print_isa_bits_for): New function.
4418 (gen_isa): New function.
4419 (gen_comm_data): Use print_isa_bits_for.
4420 (define feature): New keyword.
4421 (define fgroup): New keyword.
4422 * config/arm/t-arm (TM_H): Remove.
4423 (GTM_H): Add arm-isa.h.
4424 (arm-isa.h): Add rule to generate file.
4425 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4426 case for feature bit names.
4428 2017-09-22 Richard Biener <rguenther@suse.de>
4430 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
4432 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
4433 print of no dependency loops ...
4434 * graphite.c (graphite_transform_loops): ... here.
4435 (canonicalize_loop_closed_ssa_form): Work from inner to outer
4437 (same_close_phi_node, remove_duplicate_close_phi,
4438 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
4439 (canonicalize_loop_closed_ssa): ... here and simplify.
4440 * graphite-optimize-isl.c: Include tree-vectorizer.h.
4441 (optimize_isl): Use dump_printf_loc to tell when we stopped
4442 optimizing because of an ISL timeout.
4444 2017-09-22 Richard Biener <rguenther@suse.de>
4446 PR tree-optimization/82291
4447 * tree-if-conv.c (predicate_mem_writes): Make sure to
4448 remove writes in blocks predicated with false.
4450 2017-09-22 Richard Biener <rguenther@suse.de>
4452 * sese.c: Include cfganal.h.
4453 (if_region_set_false_region): Remove.
4454 (create_if_region_on_edge): Likewise.
4455 (move_sese_in_condition): Re-implement without destroying
4458 2017-09-22 Richard Biener <rguenther@suse.de>
4460 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
4461 Verify both BBs contain loop PHI nodes before dispatching to
4463 (graphite_regenerate_ast_isl): Do not recompute dominators,
4464 do not verify three times. Restructure for clarity.
4465 * graphite-scop-detection.c (same_close_phi_node,
4466 remove_duplicate_close_phi, make_close_phi_nodes_unique,
4467 defined_in_loop_p, canonicalize_loop_closed_ssa,
4468 canonicalize_loop_closed_ssa_form): Simplify, remove excess
4469 checking and SSA rewrite, move to ...
4470 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
4471 (graphite_initialize): Do not pass in ctx, do not reset the
4472 SCEV cache, compute only dominators.
4473 (graphite_transform_loops): Allocate ISL ctx after
4474 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
4475 Maintain post-dominators only around build_scops.
4476 * sese.c (if_region_set_false_region): Make static. Free
4477 and recompute dominators.
4478 (move_sese_in_condition): Assert we don't get called with
4479 post-dominators computed.
4480 * sese.h (if_region_set_false_region): Remove.
4482 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
4484 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
4485 mode attribute for TARGET_AVX512VL.
4487 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
4489 * config/i386/i386.opt (mprefer-avx256): New option.
4490 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
4492 (ix86_preferred_simd_mode): Return 256-bit AVX modes
4493 for TARGET_PREFER_AVX256.
4494 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
4496 2017-09-21 Jeff Law <law@redhat.com>
4498 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
4499 Fix dump output if the only stack space is for pushed registers.
4501 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4503 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
4506 2017-09-21 Martin Sebor <msebor@redhat.com>
4509 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
4510 code (in C++) or code that triggers warnings.
4512 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
4514 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
4516 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
4518 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
4519 * sched-rgn.c: Ditto.
4520 * sel-sched-ir.c: Ditto.
4522 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
4524 * toplev.h (set_random_seed): Adjust return type.
4525 * toplev.c (init_local_tick): Move eager initialization of random_seed
4526 to get_random_seed. Adjust comment.
4527 (init_random_seed): Inline to get_random_seed, delete.
4528 (get_random_seed): Initialize random_seed lazily.
4529 (set_random_seed): Do not return previous value.
4530 (print_switch_value): Do not call get_random_seed.
4532 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
4534 * cgraph.c (delete_function_version): New, broken out from...
4535 (cgraph_node::delete_function_version): ...here. Rename to
4536 cgraph_node::delete_function_version_by_decl. Update all uses.
4537 (cgraph_node::remove): Call delete_function_version.
4539 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4542 * tree-inline.c (expand_call_inline): Emit clobber stmts for
4543 VAR_DECLs to which addressable non-volatile parameters are mapped
4544 and for id->retvar after the return value assignment. Clear
4545 id->retval and id->retbnd after inlining.
4547 2017-09-21 Richard Biener <rguenther@suse.de>
4549 PR tree-optimization/82276
4550 PR tree-optimization/82244
4551 * tree-vrp.c (build_assert_expr_for): Set
4552 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
4554 (remove_range_assertions): Revert earlier change.
4556 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
4559 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
4561 2017-09-21 Richard Biener <rguenther@suse.de>
4563 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4564 Restore valid IL after code generation errors.
4565 * graphite.c (graphite_transform_loops): Diagnose code
4566 generation issues as MSG_MISSED_OPTIMIZATION and continue
4567 with processing SCOPs.
4569 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4570 Alan Hayward <alan.hayward@arm.com>
4571 David Sherwood <david.sherwood@arm.com>
4573 * calls.c (compute_argument_addresses): Use simplify_gen_binary
4574 rather than choosing between plus_constant and gen_rtx_<CODE>.
4575 * expr.c (emit_push_insn): Likewise.
4576 (expand_expr_real_2): Likewise.
4578 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4579 Alan Hayward <alan.hayward@arm.com>
4580 David Sherwood <david.sherwood@arm.com>
4582 * loop-unroll.c (split_iv): Call copy_rtx on the step.
4584 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4585 Alan Hayward <alan.hayward@arm.com>
4586 David Sherwood <david.sherwood@arm.com>
4588 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
4589 calling tree_to_uhwi.
4591 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4592 Alan Hayward <alan.hayward@arm.com>
4593 David Sherwood <david.sherwood@arm.com>
4595 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
4596 INTEGER_CST rather than a negative test for ADDR_EXPR.
4598 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4599 Alan Hayward <alan.hayward@arm.com>
4600 David Sherwood <david.sherwood@arm.com>
4602 * tree-vrp.c (extract_range_from_binary_expr_1): Check
4603 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
4605 2017-09-21 Richard Biener <rguenther@suse.de>
4607 PR tree-optimization/71351
4608 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
4609 graphite_create_new_loop_guard): Remove, fold remaining parts
4611 (translate_isl_ast_node_for): ... here and simplify.
4613 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4616 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
4617 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
4618 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
4619 alternative always use QI mode, for -Os imov (=R,R) alternative
4620 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
4623 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4624 Jeff Law <law@redhat.com>
4626 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
4627 (allocate_stack_space): New function, partially extracted from
4629 (s390_emit_prologue): Track offset to most recent stack probe.
4630 Code to allocate space moved into allocate_stack_space.
4631 Dump actions when no stack is allocated.
4632 (s390_prologue_plus_offset): New function.
4633 (s390_emit_stack_probe): Likewise.
4635 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
4637 * common.opt (Wa, Wl, Wp, g, gz=): Add
4639 (gno-column-info): Remove.
4640 (gcolumn-info): Drop RejectNegative.
4642 (gno-record-gcc-switches): Remove.
4643 (grecord-gcc-switches): Drop RejectNegative.
4644 (gno-split-dwarf): Remove.
4645 (gsplit-dwarf): Drop RejectNegative.
4646 (gno-strict-dwarf): Remove.
4647 (gstrict-dwarf): Drop RejectNegative.
4648 * config/darwin.opt (gfull, gused): Add RejectNegative.
4649 * dwarf2out.c (gen_producer_string): Drop
4650 gno-record-gcc-switches handler.
4651 * optc-gen.awk: Add g to prefixes with negative forms.
4652 * opts-common.c (remapping_prefix_p): New.
4653 (find_opt): Check it.
4654 (generate_canonical_option): Test g prefix.
4655 (option_map): Add -gno- mapping.
4656 (add_misspelling_candidates): Check remapping_prefix_p.
4658 2017-09-20 Jeff Law <law@redhat.com>
4660 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
4661 thinko in stack clash protection support.
4663 * explow.c (compute_stack_clash_protection_loop_data): Use
4664 CONST_INT_P instead of explicit test. Verify object is a
4665 CONST_INT_P before looking at INTVAL.
4666 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
4667 instead of explicit test.
4669 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4672 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
4673 address instead of to r1 and r11.
4675 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
4677 * config.gcc: Support "knm".
4678 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
4679 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4681 * config/i386/i386.c (m_KNM): Define.
4682 (processor_target_table): Add "knm".
4684 (ix86_option_override_internal): Add "knm".
4685 (ix86_issue_rate): Add PROCESSOR_KNM.
4686 (ix86_adjust_cost): Ditto.
4687 (ia32_multipass_dfa_lookahead): Ditto.
4688 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
4689 (fold_builtin_cpu): Add M_INTEL_KNM.
4690 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
4691 (processor_type): Add PROCESSOR_KNM.
4692 * config/i386/x86-tune.def: Add m_KNM.
4693 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
4695 2017-09-20 Richard Biener <rguenther@suse.de>
4697 PR tree-optimization/80213
4698 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
4699 are allowed in empty BBs as well.
4700 (canonicalize_loop_closed_ssa): Also look for other complex
4702 (scop_detection::get_sese): Include the loop-closed PHI block
4704 (scop_detection::merge_sese): Remove code adding extra blocks.
4705 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
4706 (build_scops): Assert the final returned scop is invalid.
4708 2017-09-20 Richard Biener <rguenther@suse.de>
4710 PR tree-optimization/82264
4711 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
4713 (vn_phi_lookup): Likewise.
4714 (vn_phi_insert): Likewise.
4716 2017-09-20 Jakub Jelinek <jakub@redhat.com>
4718 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
4719 that fits into uhwi or shwi, add DW_AT_const_value regardless
4720 of early_dwarf without going through RTL, using add_AT_unsigned
4723 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
4724 (DEBUG_LTO_ABBREV_SECTION): Likewise.
4725 (DEBUG_LTO_MACINFO_SECTION): Likewise.
4726 (DEBUG_MACRO_SECTION): Likewise.
4727 (DEBUG_LTO_MACRO_SECTION): Likewise.
4728 (DEBUG_STR_DWO_SECTION): Likewise.
4729 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
4730 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
4731 (DEBUG_LTO_DWO_LINE_SECTION): Define.
4732 (DEBUG_LTO_LINE_STR_SECTION): Define.
4733 (init_sections_and_labels): Initialize debug_line_str_section
4734 variable. Initialize debug_loc_section for -gdwarf-5 to
4735 DEBUG_LOCLISTS_SECTION. Formatting fixes.
4737 2017-09-20 Richard Biener <rguenther@suse.de>
4739 * graphite-sese-to-poly.c (extract_affine): Properly handle
4740 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
4742 2017-09-20 Richard Biener <rguenther@suse.de>
4744 PR tree-optimization/81373
4745 * graphite-scop-detection.c (build_cross_bb_scalars_def):
4746 Force SESE live-out defs to be handled even if they are
4749 2017-09-19 Jeff Law <law@redhat.com>
4751 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
4752 nothing for stack adjustments with REG_STACK_CHECK.
4753 * sched-deps.c (parse_add_or_inc): Reject insns with
4754 REG_STACK_CHECK from dependency breaking.
4755 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
4756 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
4757 * reg-notes.def (STACK_CHECK): New note.
4759 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
4760 (ix86_expand_prologue): Dump stack clash info as needed.
4761 Call ix86_adjust_stack_and_probe_stack_clash as needed.
4763 * function.c (dump_stack_clash_frame_info): New function.
4764 * function.h (dump_stack_clash_frame_info): Prototype.
4765 (enum stack_clash_probes): New enum.
4767 * config/alpha/alpha.c (alpha_expand_prologue): Also check
4768 flag_stack_clash_protection.
4769 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
4770 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
4771 (arm_frame_pointer_required): Likewise.
4772 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
4773 (ia64_expand_prologue): Likewise.
4774 * config/mips/mips.c (mips_expand_prologue): Likewise.
4775 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
4776 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4777 (sparc_flat_expand_prologue): Likewise.
4778 * config/spu/spu.c (spu_expand_prologue): Likewise.
4780 * explow.c: Include "params.h".
4781 (anti_adjust_stack_and_probe_stack_clash): New function.
4782 (get_stack_check_protect): Likewise.
4783 (compute_stack_clash_protection_loop_data): Likewise.
4784 (emit_stack_clash_protection_loop_start): Likewise.
4785 (emit_stack_clash_protection_loop_end): Likewise.
4786 (allocate_dynamic_stack_space): Use get_stack_check_protect.
4787 Use anti_adjust_stack_and_probe_stack_clash.
4788 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
4789 (emit_stack_clash_protection_loop_start): Likewise.
4790 (emit_stack_clash_protection_loop_end): Likewise.
4791 * rtl.h (get_stack_check_protect): Prototype.
4792 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
4793 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
4794 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
4796 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
4798 * doc/tm.texi: Rebuilt.
4799 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4800 get_stack_check_protect.
4801 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
4802 * config/arm/arm.c (arm_expand_prologue): Likewise.
4803 (arm_frame_pointer_required): Likewise.
4804 * config/i386/i386.c (ix86_expand_prologue): Likewise.
4805 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
4806 * config/mips/mips.c (mips_expand_prologue): Likewise.
4807 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
4808 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
4809 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4810 (sparc_flat_expand_prologue): Likewise.
4812 * common.opt (-fstack-clash-protection): New option.
4813 * flag-types.h (enum stack_check_type): Note difference between
4814 -fstack-check= and -fstack-clash-protection.
4815 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
4816 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
4817 * toplev.c (process_options): Issue warnings/errors for cases
4818 not handled with -fstack-clash-protection.
4819 * doc/invoke.texi (-fstack-clash-protection): Document new option.
4820 (-fstack-check): Note additional problem with -fstack-check=generic.
4821 Note that -fstack-check is primarily for Ada and refer users
4822 to -fstack-clash-protection for stack-clash-protection.
4823 Document new params for stack clash protection.
4825 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
4827 * config/i386/i386.c (ix86_split_long_move): Do not handle
4828 address used for LEA in a special way.
4830 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
4832 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
4834 2017-09-19 Martin Sebor <msebor@redhat.com>
4837 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
4838 of incompatible types.
4840 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
4842 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4843 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
4844 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4845 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
4847 2017-09-19 Richard Biener <rguenther@suse.de>
4849 PR tree-optimization/82244
4850 * tree-vrp.c (remove_range_assertions): Do not propagate
4851 a constant to abnormals but replace the assert with a copy.
4853 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
4855 PR rtl-optimization/57878
4856 PR rtl-optimization/68988
4857 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
4858 avoidance test involving non_reload_pseudos. Move frequency test
4859 below the general fragmentation avoidance test.
4861 2017-09-19 Richard Biener <rguenther@suse.de>
4863 PR tree-optimization/69728
4864 * graphite-sese-to-poly.c (schedule_error): New global.
4865 (add_loop_schedule): Handle empty domain by failing the
4867 (build_original_schedule): Handle schedule_error.
4869 2017-09-19 Richard Biener <rguenther@suse.de>
4871 * graphite-scop-detection.c (scop_detection::can_represent_loop):
4872 Do not iterate to sibling loops but only to siblings of inner
4875 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
4878 * config/m68k/m68k.md (moveq feeding equality comparison): Check
4879 that the registers are different.
4881 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
4883 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
4884 to processor_model and "amdfam17h" to arch_names_table.
4885 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
4887 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4890 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
4892 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4893 Alan Hayward <alan.hayward@arm.com>
4894 David Sherwood <david.sherwood@arm.com>
4896 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
4898 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
4900 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
4901 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
4902 vect_schedule_slp_instance.
4903 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
4904 Update call to vect_slp_analyze_node_operations. Simplify return
4906 (vect_slp_analyze_bb_1): Update call accordingly.
4907 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
4908 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
4909 (vect_schedule_slp): Update call accordingly.
4911 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4912 Alan Hayward <alan.hayward@arm.com>
4913 David Sherwood <david.sherwood@arm.com>
4915 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
4916 with types that aren't in fact scalar.
4918 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4920 * tree-vect-slp.c (vect_record_max_nunits): New function,
4922 (vect_build_slp_tree_1): ...here.
4923 (vect_build_slp_tree_2): Call it for phis too.
4925 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4927 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
4928 to vect_get_vec_def_for_operand when getting the mask operand.
4930 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4931 Alan Hayward <alan.hayward@arm.com>
4932 David Sherwood <david.sherwood@arm.com>
4934 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
4937 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4938 Alan Hayward <alan.hayward@arm.com>
4939 David Sherwood <david.sherwood@arm.com>
4941 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
4942 calculation for vector booleans.
4944 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4945 Alan Hayward <alan.hayward@arm.com>
4946 David Sherwood <david.sherwood@arm.com>
4948 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
4950 (vect_transform_stmt): ...here.
4951 (vect_analyze_stmt): Use it instead of calling
4952 vectorizable_live_operation directly.
4954 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
4956 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
4957 non-SIMT targets in acc vector loops.
4959 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
4961 * configure.ac: Add arc and check if assembler supports gdwarf2.
4962 * configure: Regenerate.
4964 2017-09-18 Richard Biener <rguenther@suse.de>
4966 PR tree-optimization/82220
4967 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
4968 epilogue niters from the min_profitable_iters compute.
4970 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4973 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
4974 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
4975 (ix86_init_pic_reg): Revert 2017-09-01 changes.
4977 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
4980 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
4981 switching to a new text section.
4983 2017-09-18 Richard Biener <rguenther@suse.de>
4985 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
4987 (build_alias_set): Reject aliases with no access function.
4989 2017-09-18 Richard Biener <rguenther@suse.de>
4991 PR tree-optimization/79622
4992 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
4994 (build_cross_bb_scalars_use): Likewise.
4996 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
4998 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
5000 2017-09-18 Alan Modra <amodra@gmail.com>
5003 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
5004 stack_pointer_rtx for count 0. Update comments. Break up
5005 large rtl expression.
5007 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
5009 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
5010 Increase to 20 bytes.
5011 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
5012 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
5013 or avx version of the stub.
5015 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
5018 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
5019 compute the minimum stack alignment. Also update preferred stack
5020 boundary for leaf functions.
5022 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
5024 PR tree-optimization/82228
5025 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
5028 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5030 * common/config/nds32/nds32-common.c
5031 (nds32_option_optimization_table): Refine formatting.
5032 (nds32_option_optimization_table): Use -fsched-pressure and
5033 -fomit-frame-pointer for specific optimization level.
5035 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5037 * config/nds32/nds32.c: Refine formatting and comments.
5038 * config/nds32/nds32.h: Likewise.
5039 * config/nds32/nds32.md: Likewise.
5040 * config/nds32/nds32-cost.c: Likewise.
5041 * config/nds32/nds32-isr.c: Likewise.
5042 * config/nds32/nds32-md-auxiliary.c: Likewise.
5043 * config/nds32/nds32-multiple.md: Likewise.
5044 * config/nds32/nds32-predicates.c: Likewise.
5046 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
5047 Jakub Jelinek <jakub@redhat.com>
5049 Add support for -std=c++2a.
5050 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
5052 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
5054 2017-09-15 Steve Ellcey <sellcey@cavium.com>
5057 * doc/extend.texi (Common Function Attributes): Add
5058 references to ARM, AArch64, and S/390 specific attributes.
5059 (Function Specific Option Pragmas): Add AArch64 and S/390
5060 to list of back ends that support the target pragma.
5062 2017-09-15 Nathan Sidwell <nathan@acm.org>
5064 * doc/standards.texi: Fix C++17 description. Update URLs for
5067 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
5069 * common.opt (Wcast-align=strict): New warning option.
5070 * doc/invoke.texi: Document -Wcast-align=strict.
5072 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
5074 * cgraph.h (cgraph_thunk_info): Add comments.
5075 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
5076 assert for VIRTUAL_* arguments stricter.
5078 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
5080 PR tree-optimization/71026
5081 * match.pd: Move RDIV patterns from fold-const.c
5082 * fold-const.c (distribute_real_division): Removed.
5083 (fold_binary_loc): Remove calls to distribute_real_divison.
5085 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5087 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
5088 c++1z and gnu++1z as deprecated. Change other references to
5089 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
5090 Change -Wc++1z-compat to -Wc++17-compat.
5091 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
5092 * dwarf2out.c (highest_c_language): Handle C++17.
5093 (gen_compile_unit_die): Likewise.
5095 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5097 PR rtl-optimization/82192
5098 * combine.c (make_extraction): Don't look through non-paradoxical
5099 SUBREGs or TRUNCATE if pos + len is or might be bigger than
5102 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5103 Alan Hayward <alan.hayward@arm.com>
5104 David Sherwood <david.sherwood@arm.com>
5106 * target.def (function_arg_offset): New hook.
5107 * targhooks.h (default_function_arg_offset): Declare.
5108 * targhooks.c (default_function_arg_offset): New function.
5109 * function.c (locate_and_pad_parm): Use
5110 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
5111 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
5112 (TARGET_FUNCTION_ARG_OFFSET): ...this.
5113 * doc/tm.texi: Regenerate.
5114 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
5115 * config/spu/spu.c (spu_function_arg_offset): New function.
5116 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
5117 * system.h (FUNCTION_ARG_OFFSET): Poison.
5119 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5120 Alan Hayard <alan.hayward@arm.com>
5121 David Sherwood <david.sherwood@arm.com>
5123 * target.def (truly_noop_truncation): New hook.
5124 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
5125 than TRULY_NOOP_TRUNCATION.
5126 * hooks.h (hook_bool_uint_uint_true): Declare.
5127 * hooks.c (hook_bool_uint_uint_true): New function.
5128 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
5129 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
5130 * doc/tm.texi: Regenerate.
5131 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
5132 rather than TRULY_NOOP_TRUNCATION in comments.
5133 (simplify_comparison): Likewise.
5134 (record_truncated_value): Likewise.
5135 * expmed.c (extract_bit_field_1): Likewise.
5136 (extract_split_bit_field): Likewise.
5137 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
5138 instead of TRULY_NOOP_TRUNCATION.
5139 * function.c (assign_parm_setup_block): Likewise.
5140 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
5141 * rtlhooks.c: Include target.h.
5142 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
5143 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
5144 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
5145 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
5146 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
5147 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
5148 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
5149 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
5150 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
5151 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
5152 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
5153 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
5154 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
5155 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
5156 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
5157 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
5158 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
5159 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
5160 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
5161 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
5162 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
5163 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
5164 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
5165 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
5166 * config/mips/mips.c (mips_truly_noop_truncation): New function.
5167 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5168 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5169 rather than TRULY_NOOP_TRUNCATION in comments.
5170 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
5171 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
5172 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
5173 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
5174 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
5175 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
5176 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
5177 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
5178 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
5179 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
5180 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
5181 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5182 rather than TRULY_NOOP_TRUNCATION in comments.
5183 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
5184 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
5185 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
5186 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
5187 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
5188 TRULY_NOOP_TRUNCATION condition.
5189 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
5190 (TRULY_NOOP_TRUNCATION): Delete.
5191 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
5192 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
5193 * config/spu/spu.c (spu_truly_noop_truncation): New function.
5194 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5195 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
5196 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
5197 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
5198 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5199 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5200 rather than TRULY_NOOP_TRUNCATION in comments.
5201 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
5202 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
5203 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
5204 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
5205 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
5206 * system.h (TRULY_NOOP_TRUNCATION): Poison.
5208 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
5211 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
5212 (*cmp_ior): Likewise.
5213 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
5214 (*ior_scc_scc_cmp): Likewise.
5215 (*and_scc_scc): Likewise.
5216 (*and_scc_scc_cmp): Likewise.
5218 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5219 Alan Hayard <alan.hayward@arm.com>
5220 David Sherwood <david.sherwood@arm.com>
5222 * target.def (can_change_mode_class): New hook.
5223 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
5224 (hard_regno_nregs): Likewise.
5225 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
5226 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
5227 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
5228 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
5229 (LOAD_EXTEND_OP): Update accordingly.
5230 * doc/tm.texi: Regenerate.
5231 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
5232 CANNOT_CHANGE_MODE_CLASS.
5233 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
5234 (REG_CAN_CHANGE_MODE_P): ...this new macro.
5235 * combine.c (simplify_set): Update accordingly.
5236 * emit-rtl.c (validate_subreg): Likewise.
5237 * recog.c (general_operand): Likewise.
5238 * regcprop.c (mode_change_ok): Likewise.
5239 * reload1.c (choose_reload_regs): Likewise.
5240 (inherit_piecemeal_p): Likewise.
5241 * rtlanal.c (simplify_subreg_regno): Likewise.
5242 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
5243 instead of CANNOT_CHANGE_MODE_CLASS.
5244 (reload_cse_simplify_operands): Likewise.
5245 * reload.c (push_reload): Use targetm.can_change_mode_class
5246 instead of CANNOT_CHANGE_MODE_CLASS.
5247 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
5248 REG_CANNOT_CHANGE_MODE_P.
5249 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5250 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
5251 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5252 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5253 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5254 (arm_can_change_mode_class): New function.
5255 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
5256 than CANNOT_CHANGE_MODE_CLASS in comments.
5257 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5258 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
5259 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
5260 (ix86_can_change_mode_class): ...this new function, inverting the
5261 sense of the return value.
5262 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5263 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5264 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5265 (ia64_can_change_mode_class): New function.
5266 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5267 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
5268 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
5269 (m32c_can_change_mode_class): ...this new function, inverting the
5270 sense of the return value.
5271 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5272 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5273 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
5274 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
5275 (mips_can_change_mode_class): ...this new function, inverting the
5276 sense of the return value.
5277 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5278 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5279 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5280 (msp430_can_change_mode_class): New function.
5281 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5282 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
5283 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5284 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5285 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5286 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
5287 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5288 (pa_cannot_change_mode_class): Replace with...
5289 (pa_can_change_mode_class): ...this new function, inverting the
5290 sense of the return value.
5291 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
5292 than CANNOT_CHANGE_MODE_CLASS in comments.
5293 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5294 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
5295 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5296 (pdp11_cannot_change_mode_class): Replace with...
5297 (pdp11_can_change_mode_class): ...this new function, inverting the
5298 sense of the return value.
5299 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5300 * config/powerpcspe/powerpcspe-protos.h
5301 (rs6000_cannot_change_mode_class_ptr): Delete.
5302 * config/powerpcspe/powerpcspe.c
5303 (rs6000_cannot_change_mode_class_ptr): Delete.
5304 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5305 (rs6000_option_override_internal): Assign to
5306 targetm.can_change_mode_class instead of
5307 rs6000_cannot_change_mode_class_ptr.
5308 (rs6000_cannot_change_mode_class): Replace with...
5309 (rs6000_can_change_mode_class): ...this new function, inverting the
5310 sense of the return value.
5311 (rs6000_debug_cannot_change_mode_class): Replace with...
5312 (rs6000_debug_can_change_mode_class): ...this new function.
5313 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5314 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
5315 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5316 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5317 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
5319 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
5320 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5321 (rs6000_option_override_internal): Assign to
5322 targetm.can_change_mode_class instead of
5323 rs6000_cannot_change_mode_class_ptr.
5324 (rs6000_cannot_change_mode_class): Replace with...
5325 (rs6000_can_change_mode_class): ...this new function, inverting the
5326 sense of the return value.
5327 (rs6000_debug_cannot_change_mode_class): Replace with...
5328 (rs6000_debug_can_change_mode_class): ...this new function.
5329 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5330 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
5331 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
5332 (s390_can_change_mode_class): ...this new function, inverting the
5333 sense of the return value.
5334 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5335 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5336 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
5337 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5338 (sh_cannot_change_mode_class): Replace with...
5339 (sh_can_change_mode_class): ...this new function, inverting the
5340 sense of the return value.
5341 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5342 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5343 (sparc_can_change_mode_class): New function.
5344 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5345 * config/spu/spu.c (spu_can_change_mode_class): New function.
5346 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5347 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5348 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5349 (visium_can_change_mode_class): New function.
5350 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
5352 2017-09-15 Richard Biener <rguenther@suse.de>
5354 PR tree-optimization/82217
5355 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
5356 but not undefined case.
5358 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5361 * postreload.c (reload_cse_simplify_operands): Skip
5362 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
5364 2017-09-15 Richard Biener <rguenther@suse.de>
5366 PR tree-optimization/68823
5367 * graphite-scop-detection.c (build_alias_set): If we have a
5368 possible dependence check whether we can handle them by just
5369 looking at the DRs DR_ACCESS_FNs.
5370 (build_scops): If build_alias_set fails, fail the SCOP.
5372 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5374 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
5375 to support float128 built-in functions that require the ISA 3.0
5377 (BU_FLOAT128_3_HW): Likewise.
5378 (SQRTF128): Add support for the IEEE 128-bit square root and fma
5380 (FMAF128): Likewise.
5382 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
5383 support for built-in functions that need the ISA 3.0 IEEE 128-bit
5384 floating point instructions.
5385 (rs6000_invalid_builtin): Likewise.
5386 (rs6000_builtin_mask_names): Likewise.
5387 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
5388 (RS6000_BTM_FLOAT128_HW): Likewise.
5389 (RS6000_BTM_COMMON): Likewise.
5390 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
5392 * doc/extend.texi (RS/6000 built-in functions): Document the
5393 IEEE 128-bit floating point square root and fused multiply-add
5396 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
5398 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
5399 reg (r2) isn't in the set of registers defined in the prologue.
5401 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5402 Alan Hayward <alan.hayward@arm.com>
5403 David Sherwood <david.sherwood@arm.com>
5405 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
5406 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
5407 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
5408 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
5409 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
5411 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
5412 max_vectorization_factor.
5413 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
5415 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5416 Alan Hayward <alan.hayward@arm.com>
5417 David Sherwood <david.sherwood@arm.com>
5419 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
5420 (vect_worthwhile_without_simd_p): Declare.
5421 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
5422 (vectorizable_reduction): Use it.
5423 * tree-vect-stmts.c (vectorizable_shift): Likewise.
5424 (vectorizable_operation): Likewise.
5426 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5427 Alan Hayward <alan.hayward@arm.com>
5428 David Sherwood <david.sherwood@arm.com>
5430 * tree-vectorizer.h (vect_get_num_copies): New function.
5431 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
5432 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5433 (vectorizable_induction): Likewise.
5434 (vectorizable_live_operation): Likewise.
5435 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5436 (vectorizable_bswap): Likewise.
5437 (vectorizable_call): Likewise.
5438 (vectorizable_conversion): Likewise.
5439 (vectorizable_assignment): Likewise.
5440 (vectorizable_shift): Likewise.
5441 (vectorizable_operation): Likewise.
5442 (vectorizable_store): Likewise.
5443 (vectorizable_load): Likewise.
5444 (vectorizable_condition): Likewise.
5445 (vectorizable_comparison): Likewise.
5446 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
5448 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5449 Alan Hayward <alan.hayward@arm.com>
5450 David Sherwood <david.sherwood@arm.com>
5452 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
5453 of vect_init_vector.
5455 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5456 Alan Hayward <alan.hayward@arm.com>
5457 David Sherwood <david.sherwood@arm.com>
5459 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
5460 an inline wrapper that provides a location.
5461 (gimple_build_vector): Likewise.
5462 * gimple-fold.c (gimple_build_vector_from_val): New function.
5463 (gimple_build_vector): Likewise.
5464 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
5465 functions to build the initial value. Always return a gimple value.
5466 (get_initial_defs_for_reduction): Likewise. Only compute
5468 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
5469 vect_init_vector on the results from get_initial_def(s)_for_reduction.
5470 (vectorizable_induction): Use gimple_build_vector rather than
5473 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5474 Alan Hayward <alan.hayward@arm.com>
5475 David Sherwood <david.sherwood@arm.com>
5477 * target.h (vec_perm_indices): New typedef.
5478 (auto_vec_perm_indices): Likewise.
5479 * optabs-query.h: Include target.h
5480 (can_vec_perm_p): Take a vec_perm_indices *.
5481 * optabs-query.c (can_vec_perm_p): Likewise.
5482 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
5483 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5484 * tree-vect-generic.c (lower_vec_perm): Likewise.
5485 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
5486 (vect_grouped_load_supported): Likewise.
5487 (vect_shift_permute_load_chain): Likewise.
5488 (vect_permute_store_chain): Use auto_vec_perm_indices.
5489 (vect_permute_load_chain): Likewise.
5490 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
5491 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
5492 Update uses of can_vec_perm_p.
5493 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
5494 mode with a number of elements. Take a vec_perm_indices *.
5495 (vect_create_epilog_for_reduction): Update accordingly.
5496 Use auto_vec_perm_indices.
5497 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
5498 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
5499 (vect_transform_slp_perm_load): Likewise.
5500 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
5501 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
5502 (vect_gen_perm_mask_checked): Likewise.
5503 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
5504 (vect_gen_perm_mask_checked): Likewise.
5505 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
5506 (vectorizable_store): Likewise.
5507 (vectorizable_load): Likewise.
5508 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
5509 (vectorizable_bswap): Likewise.
5511 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5512 Alan Hayward <alan.hayward@arm.com>
5513 David Sherwood <david.sherwood@arm.com>
5515 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
5516 * tree.c (build_vector): Likewise.
5517 (build_vector_from_ctor): Update accordingly.
5518 (build_vector_from_val): Likewise.
5519 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
5520 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5521 * tree-vect-generic.c (add_rshift): Likewise.
5522 (expand_vector_divmod): Likewise.
5523 (optimize_vector_constructor): Likewise.
5524 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
5525 (vect_transform_slp_perm_load): Likewise.
5526 (vect_schedule_slp_instance): Likewise.
5527 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
5528 (vectorizable_call): Likewise.
5529 (vect_gen_perm_mask_any): Likewise. Add elements in order.
5530 * expmed.c (make_tree): Likewise.
5531 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
5532 a vector passed to build_vector.
5533 (fold_convert_const): Likewise.
5534 (exact_inverse): Likewise.
5535 (fold_ternary_loc): Likewise.
5536 (fold_relational_const): Likewise.
5537 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
5538 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
5539 (const_unop): Likewise. Store the reduction accumulator in a
5540 variable rather than an array.
5541 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
5542 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
5543 the new vector, rather than constructing it after the input arrays.
5544 (native_interpret_vector): Use auto_vec<tree> when building
5545 a vector passed to build_vector. Add elements in order.
5546 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
5547 auto_vec<tree> when building a vector passed to build_vector.
5548 (vect_create_epilog_for_reduction): Likewise.
5549 (vectorizable_induction): Likewise.
5550 (get_initial_def_for_reduction): Likewise. Fix indentation of
5552 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
5554 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
5555 passed to build_vector.
5557 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5558 Alan Hayward <alan.hayward@arm.com>
5559 David Sherwood <david.sherwood@arm.com>
5561 * tree-core.h (tree_base::u): Add an "nelts" field.
5562 (tree_vector): Use VECTOR_CST_NELTS as the length.
5563 * tree.c (tree_size): Likewise.
5564 (make_vector): Initialize VECTOR_CST_NELTS.
5565 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
5566 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
5567 TYPE_VECTOR_SUBPARTS.
5568 * expr.c (const_vector_mask_from_tree): Consistently use "units"
5569 as the number of units, setting it from VECTOR_CST_NELTS.
5570 (const_vector_from_tree): Likewise.
5571 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
5572 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
5573 (fold_negate_expr_1): Likewise.
5574 (fold_convert_const): Likewise.
5575 (const_binop): Likewise. Differentiate the number of output and
5577 (const_unop): Likewise.
5578 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
5579 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
5580 in cases that did the opposite.
5582 2017-09-14 Richard Biener <rguenther@suse.de>
5584 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
5587 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
5589 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
5591 2017-09-14 Jakub Jelinek <jakub@redhat.com>
5594 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
5595 if and where to split a bb, except for splitting before debug insn
5596 sequences followed by non-label real insn. Delete debug insns
5597 in between basic blocks.
5599 * combine.c (make_compound_operation_int): Formatting fixes.
5601 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
5602 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5603 * config/netbsd.h (LINK_EH_SPEC): Likewise.
5604 * config/sol2.h (LINK_EH_SPEC): Likewise.
5605 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5606 * config/s390/linux.h (LINK_SPEC): Likewise.
5607 * config/freebsd.h (LINK_EH_SPEC): Likewise.
5608 * config/openbsd.h (LINK_EH_SPEC): Likewise.
5609 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5610 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
5611 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
5612 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5613 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
5614 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
5616 2017-09-13 Jakub Jelinek <jakub@redhat.com>
5618 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
5620 (ENDFILE_LINUX_SPEC): Likewise.
5621 (LINK_EH_SPEC): Likewise.
5622 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
5623 (LINK_OS_LINUX_SPEC32): Likewise.
5624 (LINK_OS_LINUX_SPEC64): Likewise.
5625 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
5626 (LINK_OS_LINUX_SPEC): Likewise.
5628 2017-09-13 Martin Liska <mliska@suse.cz>
5631 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
5632 CASE_HIGH is NULL_TREE.
5634 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5635 Alan Hayward <alan.hayward@arm.com>
5636 David Sherwood <david.sherwood@arm.com>
5638 * target.def (secondary_memory_needed): New hook.
5639 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
5640 instead of SECONDARY_MEMORY_NEEDED.
5641 (secondary_memory_needed_mode): Likewise.
5642 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
5643 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
5644 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
5645 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
5646 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
5647 * doc/tm.texi: Regenerate.
5648 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
5649 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
5650 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5651 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
5652 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
5653 * config/i386/i386.c (inline_secondary_memory_needed): Put the
5654 mode argument first and change the reg_class arguments to reg_class_t.
5655 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
5656 Make static. Update the call to inline_secondary_memory_needed.
5657 (ix86_register_move_cost): Update the call to
5658 inline_secondary_memory_needed.
5659 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5660 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
5662 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
5663 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5665 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
5666 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
5667 * config/mips/mips.c (mips_secondary_memory_needed): Make static
5668 and match hook interface. Add comment from mips.h.
5669 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5670 * config/mmix/mmix.md (truncdfsf2): Refer to
5671 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5673 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
5674 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
5675 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5676 (pa_secondary_memory_needed): New function.
5677 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
5678 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
5679 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5680 (pdp11_secondary_memory_needed): Make static and match hook interface.
5681 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
5682 * config/powerpcspe/powerpcspe-protos.h
5683 (rs6000_secondary_memory_needed_ptr): Delete.
5684 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
5686 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5687 (rs6000_option_override_internal): Assign to
5688 targetm.secondary_memory_needed rather than
5689 rs6000_secondary_memory_needed_ptr.
5690 (rs6000_secondary_memory_needed): Match hook interface.
5691 (rs6000_debug_secondary_memory_needed): Likewise.
5692 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
5693 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
5694 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
5695 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5696 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
5697 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
5699 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
5700 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5701 (rs6000_option_override_internal): Assign to
5702 targetm.secondary_memory_needed rather than
5703 rs6000_secondary_memory_needed_ptr.
5704 (rs6000_secondary_memory_needed): Match hook interface.
5705 (rs6000_debug_secondary_memory_needed): Likewise.
5706 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
5707 * config/s390/s390.c (s390_secondary_memory_needed): New function.
5708 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5709 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
5710 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5711 (sparc_secondary_memory_needed): New function.
5712 * lra-constraints.c (check_and_process_move): Refer to
5713 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5715 (curr_insn_transform): Likewise.
5716 (process_alt_operands): Use targetm.secondary_memory_needed
5717 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5718 (check_secondary_memory_needed_p): Likewise.
5719 (choose_split_class): Likewise.
5720 * reload.c: Unconditionally include code that was previously
5721 conditional on SECONDARY_MEMORY_NEEDED.
5722 (push_secondary_reload): Use targetm.secondary_memory_needed
5723 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5724 (push_reload): Likewise.
5725 * reload1.c: Unconditionally include code that was previously
5726 conditional on SECONDARY_MEMORY_NEEDED.
5727 (choose_reload_regs): Use targetm.secondary_memory_needed
5728 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5729 (gen_reload): Likewise.
5730 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
5732 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5733 Alan Hayward <alan.hayward@arm.com>
5734 David Sherwood <david.sherwood@arm.com>
5736 * target.def (secondary_memory_needed_mode): New hook:
5737 * targhooks.c (default_secondary_memory_needed_mode): Declare.
5738 * targhooks.h (default_secondary_memory_needed_mode): New function.
5739 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
5740 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
5741 * doc/tm.texi: Regenerate.
5742 * lra-constraints.c (check_and_process_move): Use
5743 targetm.secondary_memory_needed_mode instead of
5744 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
5745 (curr_insn_transform): Likewise.
5746 * reload.c (get_secondary_mem): Likewise.
5747 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5748 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
5750 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5751 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5752 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
5753 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5754 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
5756 * config/powerpcspe/powerpcspe-protos.h
5757 (rs6000_secondary_memory_needed_mode): Delete.
5758 * config/powerpcspe/powerpcspe.c
5759 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5760 (rs6000_secondary_memory_needed_mode): Make static.
5761 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5762 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
5764 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5766 (rs6000_secondary_memory_needed_mode): Make static.
5767 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5768 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
5769 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5770 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5771 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5773 (sparc_secondary_memory_needed_mode): New function.
5774 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
5776 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
5778 * config/aarch64/constraints.md (Umq): New constraint.
5779 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
5781 (mov<mode>): Update condition.
5783 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5785 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
5786 when bitposition is the same.
5788 2017-09-13 Richard Biener <rguenther@suse.de>
5790 * dwarf2out.c (output_die_symbol): Remove.
5791 (output_die): Do not output a DIEs symbol.
5793 2017-09-13 Richard Biener <rguenther@suse.de>
5796 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
5797 default-def to avoid breaking iterator update with the weird
5798 interaction with cgraph_update_edges_for_call_stmt_node.
5800 2017-09-13 Richard Biener <rguenther@suse.de>
5802 * tree-cfg.c (verify_gimple_assign_binary): Add verification
5803 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
5804 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
5805 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
5807 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
5809 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
5810 Disable pc relative literal load irrespective of
5811 TARGET_FIX_ERR_A53_84341 for default.
5813 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
5815 * config/sparc/sparc.c (output_return): Output the source location of
5816 the insn in the delay slot, if any.
5817 (output_sibcall): Likewise.
5819 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
5822 * common.opt (-static-pie): New alias.
5823 (shared): Negate static-pie.
5824 (-no-pie): Update help text.
5826 (static-pie): New option.
5827 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
5828 -static-pie support.
5829 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
5830 (LINK_EH_SPEC): Likewise.
5831 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5832 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5833 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5834 * gcc.c (LINK_COMMAND_SPEC): Likewise.
5835 (init_gcc_specs): Likewise.
5836 (init_spec): Likewise.
5837 (display_help): Update help message for -pie.
5838 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
5841 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
5843 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
5844 (movdi_aarch64): Likewise.
5845 (movti_aarch64): Likewise.
5847 2017-09-12 Simon Wright <simon@pushface.org>
5850 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
5851 calculation of the minor version, always output as 0.
5853 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5856 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5857 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
5858 on it early, rather than manual conversion late. For
5859 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
5860 instead of performing manual conversion.
5862 2017-09-12 Carl Love <cel@us.ibm.com>
5864 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
5865 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
5867 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5868 VMULOSW): Add definitions.
5869 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5870 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5871 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5872 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5873 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5875 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5877 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
5879 (movti_aarch64): Likewise.
5880 (movdf_aarch64): Likewise.
5881 (movtf_aarch64): Likewise.
5882 (load_pairdi): Likewise.
5883 (store_pairdi): Likewise.
5884 (load_pairdf): Likewise.
5885 (store_pairdf): Likewise.
5886 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
5887 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
5888 (ldr_got_small_<mode>): Likewise.
5889 (ldr_got_small_28k_<mode>): Likewise.
5890 (ldr_got_tiny): Likewise.
5891 * config/aarch64/iterators.md (ldst_sz): New.
5892 (ldpstp_sz): Likewise.
5893 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
5895 (thunderx_load): Split load_8 to load_16.
5896 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
5898 (thunderx2t99_storepair_basic): Split store_8 to store_16.
5899 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
5900 (xgene1_store_pair): Split store_8 to store_16.
5901 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
5902 (falkor_st_0_st_sd): Split store_8 to store_16.
5904 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5906 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
5907 and store1/2/3/4 to store_4/8/12/16.
5908 * config/aarch64/aarch64.md: Update for rename.
5909 * config/arm/arm.md: Likewise.: Likewise.
5910 * config/arm/arm.c: Likewise.
5911 * config/arm/thumb1.md: Likewise.
5912 * config/arm/thumb2.md: Likewise.
5913 * config/arm/vfp.md: Likewise.
5914 * config/arm/arm-generic.md: Likewise.
5915 * config/arm/arm1020e.md: Likewise.
5916 * config/arm/arm1026ejs.md: Likewise.
5917 * config/arm/arm1136jfs.md: Likewise.
5918 * config/arm/arm926ejs.md: Likewise.
5919 * config/arm/cortex-a15.md: Likewise.
5920 * config/arm/cortex-a17.md: Likewise.
5921 * config/arm/cortex-a5.md: Likewise.
5922 * config/arm/cortex-a53.md: Likewise.
5923 * config/arm/cortex-a57.md: Likewise.
5924 * config/arm/cortex-a7.md: Likewise.
5925 * config/arm/cortex-a8.md: Likewise.
5926 * config/arm/cortex-a9.md: Likewise.
5927 * config/arm/cortex-m4.md: Likewise.
5928 * config/arm/cortex-m7.md: Likewise.
5929 * config/arm/cortex-r4.md: Likewise.
5930 * config/arm/exynos-m1.md: Likewise.
5931 * config/arm/fa526.md: Likewise.
5932 * config/arm/fa606te.md: Likewise.
5933 * config/arm/fa626te.md: Likewise.
5934 * config/arm/fa726te.md: Likewise.
5935 * config/arm/fmp626.md: Likewise.
5936 * config/arm/iwmmxt.md: Likewise.
5937 * config/arm/ldmstm.md: Likewise.
5938 * config/arm/marvell-pj4.md: Likewise.
5939 * config/arm/xgene1.md: Likewise.
5940 * config/aarch64/thunderx.md: Likewise.
5941 * config/aarch64/thunderx2t99.md: Likewise.
5942 * config/aarch64/falkor.md: Likewise.
5944 2017-09-12 Martin Liska <mliska@suse.cz>
5946 * attribs.c (private_lookup_attribute): New function.
5947 * attribs.h (private_lookup_attribute): Declared here.
5948 (lookup_attribute): Called from this place.
5950 2017-09-12 Richard Biener <rguenther@suse.de>
5952 PR tree-optimization/82157
5953 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
5954 stmts with side-effects.
5956 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5957 Alan Hayward <alan.hayward@arm.com>
5958 David Sherwood <david.sherwood@arm.com>
5960 * target.def (hard_regno_nregs): New hook.
5961 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
5962 * targhooks.h (default_hard_regno_nregs): Declare.
5963 * targhooks.c (default_hard_regno_nregs): New function.
5964 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
5965 (TARGET_HARD_REGNO_NREGS): ...this hook.
5966 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
5967 (CLASS_MAX_NREGS): Likewise.
5968 * doc/tm.texi: Regenerate.
5969 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
5970 instead of HARD_REGNO_NREGS.
5971 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
5972 HARD_REGNO_NREGS in the comment.
5973 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
5974 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
5975 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
5976 Return an unsigned int.
5977 (TARGET_HARD_REGNO_NREGS): Redefine.
5978 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
5979 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
5980 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
5981 (arc_hard_regno_nregs): New function.
5982 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
5983 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
5984 (arm_hard_regno_nregs): New function.
5985 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
5986 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
5987 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
5988 (TARGET_HARD_REGNO_NREGS): Redefine.
5989 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
5990 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
5991 (HARD_REGNO_NREGS): Delete.
5992 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
5993 (cr16_hard_regno_nregs): New function.
5994 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
5995 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
5996 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
5997 (cris_hard_regno_nregs): New function.
5998 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
5999 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
6000 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6001 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
6002 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
6003 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
6004 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
6005 (frv_hard_regno_nregs): Make static. Take and return an
6007 (frv_class_max_nregs): Remove outdated copy of documentation.
6008 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
6009 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
6010 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
6011 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
6012 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
6013 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
6014 (TARGET_HARD_REGNO_NREGS): Redefine.
6015 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
6016 (CLASS_MAX_NREGS): Update comment.
6017 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
6018 (ia64_hard_regno_nregs): New function.
6019 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
6020 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
6021 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
6022 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
6023 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
6025 (m32c_hard_regno_nregs): Likewise. Make static.
6026 (TARGET_HARD_REGNO_NREGS): Redefine.
6027 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
6028 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
6029 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
6030 (m68k_hard_regno_nregs): New function.
6031 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
6032 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
6033 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
6034 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
6035 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
6036 Take and return an unsigned int.
6037 (TARGET_HARD_REGNO_NREGS): Redefine.
6038 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
6039 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6040 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
6041 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
6042 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
6043 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
6044 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
6045 (msp430_hard_regno_nregs): Make static. Take and return an
6047 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
6048 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
6049 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
6050 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
6051 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
6052 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
6053 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
6054 (TARGET_HARD_REGNO_NREGS): Redefine.
6055 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
6056 (PA_HARD_REGNO_NREGS): ...this.
6057 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
6058 (PA_HARD_REGNO_NREGS): ...this.
6059 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6060 (pa_hard_regno_nregs): New function.
6061 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
6062 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
6063 (pdp11_hard_regno_nregs): New function.
6064 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
6065 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
6066 (rs6000_hard_regno_nregs_hook): New function.
6067 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
6068 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6069 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
6070 Take and return an unsigned int. Move earlier in file.
6071 (TARGET_HARD_REGNO_NREGS): Redefine.
6072 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
6073 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
6074 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
6075 (rl78_hard_regno_nregs): Make static. Take and return an
6077 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
6078 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
6079 (rs6000_hard_regno_nregs_hook): New function.
6080 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
6081 * config/rx/rx.c (rx_hard_regno_nregs): New function.
6082 (TARGET_HARD_REGNO_NREGS): Redefine.
6083 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
6084 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
6085 instead of HARD_REGNO_NREGS.
6086 (s390_hard_regno_nregs): New function.
6087 (s390_hard_regno_mode_ok): Add comment from s390.h.
6088 (TARGET_HARD_REGNO_NREGS): Redefine.
6089 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
6090 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
6091 (sh_hard_regno_nregs): New function.
6092 (sh_pass_in_reg_p): Use it.
6093 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
6094 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6095 (sparc_hard_regno_nregs): New function.
6096 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
6097 * config/spu/spu.c (spu_hard_regno_nregs): New function.
6098 (spu_function_arg_advance): Use it, supplying a valid register number.
6099 (TARGET_HARD_REGNO_NREGS): Redefine.
6100 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
6101 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
6102 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
6103 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
6104 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
6105 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
6106 (CLASS_MAX_NREGS): Remove copy of old documentation.
6107 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
6108 (visium_hard_regno_nregs): New function.
6109 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
6110 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
6111 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6112 xtensa_hard_regno_nregs): New function.
6113 * system.h (HARD_REGNO_NREGS): Poison.
6115 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6117 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
6118 hard_regno_nregs instead of HARD_REGNO_NREGS.
6119 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
6120 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
6121 (c6x_expand_epilogue): Likewise.
6122 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
6123 (frv_read_iacc_argument): Likewise.
6124 * config/sh/sh.c: Include regs.h.
6125 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
6126 (regs_used): Likewise.
6127 (output_stack_adjust): Likewise.
6128 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
6129 * expmed.c: Include regs.h.
6130 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
6131 * ree.c: Include regs.h.
6132 (combine_reaching_defs): Use hard_regno_nregs instead of
6134 (add_removable_extension): Likewise.
6136 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6138 * regs.h (hard_regno_nregs): Turn into a function.
6139 (end_hard_regno): Update accordingly.
6140 * caller-save.c (setup_save_areas): Likewise.
6141 (save_call_clobbered_regs): Likewise.
6142 (replace_reg_with_saved_mem): Likewise.
6143 (insert_restore): Likewise.
6144 (insert_save): Likewise.
6145 * combine.c (can_change_dest_mode): Likewise.
6146 (move_deaths): Likewise.
6147 (distribute_notes): Likewise.
6148 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
6149 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
6150 (rs6000_split_multireg_move): Likewise.
6151 (rs6000_register_move_cost): Likewise.
6152 (rs6000_memory_move_cost): Likewise.
6153 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
6154 (rs6000_split_multireg_move): Likewise.
6155 (rs6000_register_move_cost): Likewise.
6156 (rs6000_memory_move_cost): Likewise.
6157 * cselib.c (cselib_reset_table): Likewise.
6158 (cselib_lookup_1): Likewise.
6159 * emit-rtl.c (set_mode_and_regno): Likewise.
6160 * function.c (aggregate_value_p): Likewise.
6161 * ira-color.c (setup_profitable_hard_regs): Likewise.
6162 (check_hard_reg_p): Likewise.
6163 (calculate_saved_nregs): Likewise.
6164 (assign_hard_reg): Likewise.
6165 (improve_allocation): Likewise.
6166 (calculate_spill_cost): Likewise.
6167 * ira-emit.c (modify_move_list): Likewise.
6168 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
6169 (ira_hard_reg_in_set_p): Likewise.
6170 * ira.c (setup_reg_mode_hard_regset): Likewise.
6171 (clarify_prohibited_class_mode_regs): Likewise.
6172 (check_allocation): Likewise.
6173 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6174 (lra_setup_reg_renumber): Likewise.
6175 (setup_try_hard_regno_pseudos): Likewise.
6176 (spill_for): Likewise.
6177 (assign_hard_regno): Likewise.
6178 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
6179 * lra-constraints.c (in_class_p): Likewise.
6180 (lra_constraint_offset): Likewise.
6181 (simplify_operand_subreg): Likewise.
6182 (lra_constraints): Likewise.
6183 (split_reg): Likewise.
6184 (split_if_necessary): Likewise.
6185 (invariant_p): Likewise.
6186 (inherit_in_ebb): Likewise.
6187 * lra-lives.c (process_bb_lives): Likewise.
6188 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
6189 (get_hard_regs): Likewise.
6190 (do_remat): Likewise.
6191 * lra-spills.c (assign_spill_hard_regs): Likewise.
6192 * mode-switching.c (create_pre_exit): Likewise.
6193 * postreload.c (reload_combine_recognize_pattern): Likewise.
6194 * recog.c (peep2_find_free_register): Likewise.
6195 * regcprop.c (kill_value_regno): Likewise.
6196 (set_value_regno): Likewise.
6197 (copy_value): Likewise.
6198 (maybe_mode_change): Likewise.
6199 (find_oldest_value_reg): Likewise.
6200 (copyprop_hardreg_forward_1): Likewise.
6201 * regrename.c (check_new_reg_p): Likewise.
6202 (regrename_do_replace): Likewise.
6203 * reload.c (push_reload): Likewise.
6204 (combine_reloads): Likewise.
6205 (find_dummy_reload): Likewise.
6206 (operands_match_p): Likewise.
6207 (find_reloads): Likewise.
6208 (find_equiv_reg): Likewise.
6209 (reload_adjust_reg_for_mode): Likewise.
6210 * reload1.c (count_pseudo): Likewise.
6211 (count_spilled_pseudo): Likewise.
6212 (find_reg): Likewise.
6213 (clear_reload_reg_in_use): Likewise.
6214 (free_for_value_p): Likewise.
6215 (allocate_reload_reg): Likewise.
6216 (choose_reload_regs): Likewise.
6217 (reload_adjust_reg_for_temp): Likewise.
6218 (emit_reload_insns): Likewise.
6219 (delete_output_reload): Likewise.
6220 * rtlanal.c (subreg_get_info): Likewise.
6221 * sched-deps.c (sched_analyze_reg): Likewise.
6222 * sel-sched.c (init_regs_for_mode): Likewise.
6223 (mark_unavailable_hard_regs): Likewise.
6224 (choose_best_reg_1): Likewise.
6225 (verify_target_availability): Likewise.
6226 * valtrack.c (dead_debug_insert_temp): Likewise.
6227 * var-tracking.c (track_loc_p): Likewise.
6228 (emit_note_insn_var_location): Likewise.
6229 * varasm.c (make_decl_rtl): Likewise.
6230 * reginfo.c (choose_hard_reg_mode): Likewise.
6231 (init_reg_modes_target): Refer directly to
6232 this_target_regs->x_hard_regno_nregs.
6234 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6236 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
6237 instead of hard_regno_nregs.
6239 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6241 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
6242 end_hard_regno instead of hard_regno_nregs.
6243 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
6244 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
6245 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
6246 * ira-color.c (improve_allocation): Likewise.
6247 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6248 * lra-lives.c (mark_regno_live): Likewise.
6249 (mark_regno_dead): Likewise.
6250 * lra-remat.c (operand_to_remat): Likewise.
6251 * lra.c (collect_non_operand_hard_regs): Likewise.
6252 * postreload.c (reload_combine_note_store): Likewise.
6253 (move2add_valid_value_p): Likewise.
6254 * reload.c (regno_clobbered_p): Likewise.
6256 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6258 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
6260 * config/v850/v850.c (v850_reorg): Likewise.
6261 * reload.c (refers_to_regno_for_reload_p): Likewise.
6262 (find_equiv_reg): Likewise.
6263 * reload1.c (reload_reg_reaches_end_p): Likewise.
6265 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6267 * caller-save.c (add_used_regs): Use REG_NREGS instead of
6269 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
6270 * config/arm/arm.c (output_move_neon): Likewise.
6271 (arm_attr_length_move_neon): Likewise.
6272 (neon_split_vcombine): Likewise.
6273 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
6274 (c6x_mark_reg_written): Likewise.
6275 (c6x_dwarf_register_span): Likewise.
6276 * config/i386/i386.c (ix86_save_reg): Likewise.
6277 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
6278 (rws_access_reg): Likewise.
6279 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
6280 * mode-switching.c (create_pre_exit): Likewise.
6281 * ree.c (combine_reaching_defs): Likewise.
6282 (add_removable_extension): Likewise.
6283 * regcprop.c (find_oldest_value_reg): Likewise.
6284 (copyprop_hardreg_forward_1): Likewise.
6285 * reload.c (reload_inner_reg_of_subreg): Likewise.
6286 (push_reload): Likewise.
6287 (combine_reloads): Likewise.
6288 (find_dummy_reload): Likewise.
6289 (reload_adjust_reg_for_mode): Likewise.
6290 * reload1.c (find_reload_regs): Likewise.
6291 (forget_old_reloads_1): Likewise.
6292 (reload_reg_free_for_value_p): Likewise.
6293 (reload_adjust_reg_for_temp): Likewise.
6294 (emit_reload_insns): Likewise.
6295 (delete_output_reload): Likewise.
6296 * sel-sched.c (choose_best_reg_1): Likewise.
6297 (choose_best_pseudo_reg): Likewise.
6299 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6300 Alan Hayward <alan.hayward@arm.com>
6301 David Sherwood <david.sherwood@arm.com>
6303 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
6304 * target.def (slow_unaligned_access): New hook.
6305 * targhooks.h (default_slow_unaligned_access): Declare.
6306 * targhooks.c (default_slow_unaligned_access): New function.
6307 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
6308 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
6309 * doc/tm.texi: Regenerate.
6310 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
6311 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
6312 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
6314 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
6315 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
6317 (rs6000_slow_unaligned_access): New function.
6318 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
6319 (expand_block_compare): Likewise.
6320 (expand_strn_compare): Likewise.
6321 (rs6000_rtx_costs): Likewise.
6322 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
6323 (riscv_slow_unaligned_access): Likewise.
6324 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
6325 (riscv_slow_unaligned_access_p): ...this and make static.
6326 (riscv_option_override): Update accordingly.
6327 (riscv_slow_unaligned_access): New function.
6328 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
6329 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
6330 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
6331 (rs6000_slow_unaligned_access): New function.
6332 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
6333 (rs6000_rtx_costs): Likewise.
6334 * config/rs6000/rs6000-string.c (expand_block_compare)
6335 (expand_strn_compare): Use targetm.slow_unaligned_access instead
6336 of SLOW_UNALIGNED_ACCESS.
6337 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
6338 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
6339 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
6340 of SLOW_UNALIGNED_ACCESS.
6341 * expmed.c (simple_mem_bitfield_p): Likewise.
6342 * expr.c (alignment_for_piecewise_move): Likewise.
6343 (emit_group_load_1): Likewise.
6344 (emit_group_store): Likewise.
6345 (copy_blkmode_from_reg): Likewise.
6346 (emit_push_insn): Likewise.
6347 (expand_assignment): Likewise.
6348 (store_field): Likewise.
6349 (expand_expr_real_1): Likewise.
6350 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
6351 * lra-constraints.c (simplify_operand_subreg): Likewise.
6352 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6353 * gimple-ssa-store-merging.c: Likewise in block comment at start
6355 * tree-ssa-strlen.c: Include target.h.
6356 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
6357 of SLOW_UNALIGNED_ACCESS.
6358 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
6360 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6362 PR rtl-optimization/82185
6363 * expmed.c (emit_store_flag_int): Only test tem if it has been
6366 2017-09-12 Richard Biener <rguenther@suse.de>
6369 * match.pd ((FTYPE) N CMP CST): Fix typo.
6371 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
6373 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
6375 (mips_near_type_p): Add 'short_call' attribute as a synonym
6377 * doc/extend.texi (short_call): Document new function attribute.
6379 2017-09-12 Jakub Jelinek <jakub@redhat.com>
6382 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
6383 assertion check that in the condition.
6384 (get_atomic_generic_size): Likewise. Before testing if parameter
6385 has pointer type, if it has array type, call for C++
6386 default_conversion to perform array-to-pointer conversion.
6388 2017-09-12 Richard Biener <rguenther@suse.de>
6390 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
6391 for operations we cannot scalarize.
6393 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
6395 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
6396 vectors heap vectors. Clean up comments.
6397 Make visited_bbs a reference.
6398 (profitable_jump_thread_path): Make GC
6399 vectors heap vectors. Clean up comments.
6401 (convert_and_register_jump_thread_path): Make GC vectors heap
6403 (check_subpath_and_update_thread_path): Same. Clean up comments.
6404 Make visited_bbs a reference.
6405 (handle_phi): Abstract common code to to
6406 register_jump_thread_path_if_profitable.
6407 Rename VAR_BB to DEF_BB.
6409 Make GC vectors heap vectors.
6410 Make visited_bbs a reference.
6411 (handle_assignment): Same.
6412 (register_jump_thread_path_if_profitable): New.
6413 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
6415 Make GC vectors heap vectors. Clean up comments.
6416 Make visited_bbs a reference.
6417 (find_jump_threads_backwards): Make visited_bbs live in the stack.
6418 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
6421 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
6424 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
6425 words of E_DImode object are reachable by xtensa_uimm8x4 access.
6427 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
6429 Revert r251800 and r251799.
6431 2017-09-11 Martin Jambor <mjambor@suse.cz>
6434 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
6435 arguments in advance.
6436 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
6437 use it to find predecessor edges.
6438 (naive_outof_ssa): Collect vector of predecessors.
6440 2017-09-08 Jason Merrill <jason@redhat.com>
6442 PR c++/70029 - ICE with ref-qualifier and -flto
6443 * langhooks.h (struct lang_hooks_for_types): Add
6444 copy_lang_qualifiers.
6445 * attribs.c (build_type_attribute_qual_variant): Use it.
6446 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
6448 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
6449 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
6451 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
6454 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
6455 (*mulsi3_sp64): New instruction.
6456 (mulsi3): New expander.
6458 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
6460 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
6462 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6464 * sancov.c: Include memmodel.h.
6466 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
6469 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
6472 2017-09-07 Carl Love <cel@us.ibm.com>
6474 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
6475 the sldi instruction.
6477 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
6479 * sancov.c: Include tm_p.h.
6481 2017-09-07 Jakub Jelinek <jakub@redhat.com>
6484 * output.h (switch_to_other_text_partition): New declaration.
6485 * varasm.c (switch_to_other_text_partition): New function.
6486 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
6487 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
6488 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
6489 to the other text partition before emitting LCL label and switch back
6490 after emitting the word after it.
6492 2017-09-07 Richard Biener <rguenther@suse.de>
6494 * passes.def (pass_split_crit_edges): Remove instance before PRE.
6495 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
6496 critical edges here, after loop init.
6497 (pass_data_pre): Remove PROP_no_crit_edges flags.
6498 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
6499 for valueization of call args to avoid leaking VN_TOP.
6500 (visit_use): Assert we do not visit default defs.
6501 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
6502 Use error_mark_node to more easily detect leaking VN_TOP.
6503 All default-defs are varying, not VN_TOP. Mark them visited.
6504 (run_scc_vn): Make code match comment.
6506 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6508 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
6509 OPTION_MASK_FLOAT128_KEYWORD.
6510 (POWERPC_MASKS): Likewise.
6511 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
6512 support for the -mfloat128-type option, and make -mfloat128
6513 default on PowerPC Linux systems. Define or undefine
6514 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
6515 Define __float128 to be __ieee128 if IEEE 128-bit support is
6516 enabled, or undefine it.
6517 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
6518 Delete defining __FLOAT128_TYPE__.
6519 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
6520 -mfloat128-type option and make -mfloat128 default on PowerPC
6522 (TARGET_FLOAT128_TYPE): Likewise.
6523 (-mfloat128-type): Likewise.
6524 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6525 Delete the -mfloat128-type option and make -mfloat128 default on
6526 PowerPC Linux systems. Always use __ieee128 to be the keyword for
6527 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
6528 128-bit floating point is enabled. Change tests from using
6529 -mfloat128-type to -mfloat128.
6530 (rs6000_mangle_type): Use the correct mangling for the __float128
6531 type even if normal long double is restricted to 64-bits.
6532 (floatn_mode): Enable the _Float128 type by default on VSX Linux
6534 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
6535 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
6536 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
6538 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
6539 documentation for -mfloat128.
6541 2017-09-06 Olivier Hainque <hainque@adacore.com>
6543 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
6545 2017-09-06 Wish Wu <wishwu007@gmail.com>
6546 Jakub Jelinek <jakub@redhat.com>
6548 * asan.c (initialize_sanitizer_builtins): Add
6549 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
6550 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
6551 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
6552 BT_FN_VOID_UINT64_PTR variables.
6553 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
6554 (BT_FN_VOID_UINT16_UINT16): Likewise.
6555 (BT_FN_VOID_UINT32_UINT32): Likewise.
6556 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
6557 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
6558 (BT_FN_VOID_UINT64_PTR): Likewise.
6559 * common.opt (flag_sanitize_coverage): New variable.
6560 (fsanitize-coverage=trace-pc): Remove.
6561 (fsanitize-coverage=): Add.
6562 * flag-types.h (enum sanitize_coverage_code): New enum.
6563 * fold-const.c (fold_range_test): Disable non-short-circuit
6564 optimization if flag_sanitize_coverage.
6565 (fold_truth_andor): Likewise.
6566 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
6567 * opts.c (COVERAGE_SANITIZER_OPT): Define.
6568 (coverage_sanitizer_opts): New array.
6569 (get_closest_sanitizer_option): Add OPTS argument, handle also
6570 OPT_fsanitize_coverage_.
6571 (parse_sanitizer_options): Adjusted to also handle
6572 OPT_fsanitize_coverage_.
6573 (common_handle_option): Add OPT_fsanitize_coverage_.
6574 * sancov.c (instrument_comparison, instrument_switch): New function.
6575 (sancov_pass): Add trace-cmp support.
6576 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
6577 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
6578 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
6579 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
6580 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
6581 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
6582 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
6583 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
6584 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
6585 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
6587 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6589 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6590 error. Only quit immediately if parsing is complete.
6591 (BEGIN): Initialize fatal_err and parse_done.
6592 (begin fpu, end fpu): Check number of arguments.
6593 (begin arch, end arch): Likewise.
6594 (begin cpu, end cpu): Likewise.
6595 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6596 (optalias): Likewise.
6598 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6600 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6601 * config/arm/arm-isa.h: Delete. Move definitions to ...
6602 * arm-cpus.in: ... here. Use new feature and fgroup values.
6603 * config/arm/arm.c (arm_option_override): Use lower case for feature
6605 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6606 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6607 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6609 (print_isa_bits_for): New function.
6610 (gen_isa): New function.
6611 (gen_comm_data): Use print_isa_bits_for.
6612 (define feature): New keyword.
6613 (define fgroup): New keyword.
6614 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
6615 (arm-isa.h): Add rule to generate file.
6616 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6617 case for feature bit names.
6619 2017-09-06 Richard Biener <rguenther@suse.de>
6621 * tree-ssa-pre.c (NECESSARY): Remove.
6622 (create_expression_by_pieces): Do not touch pass-local flags.
6623 (insert_into_preds_of_block): Likewise.
6624 (do_pre_regular_insertion): Likewise.
6625 (eliminate_insert): Likewise.
6626 (eliminate_dom_walker::before_dom_children): Likewise.
6627 (fini_eliminate): Do not look at inserted_exprs.
6628 (mark_operand_necessary): Remove.
6629 (remove_dead_inserted_code): Replace with simple work-list
6630 algorithm based on inserted_exprs and SSA uses.
6631 (pass_pre::execute): Re-order fini_eliminate and
6632 remove_dead_inserted_code.
6634 2017-09-06 Olivier Hainque <hainque@adacore.com>
6636 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
6637 for VxWorks 7. Adjust surrounding comments.
6639 2017-09-06 Richard Biener <rguenther@suse.de>
6641 * gimple-ssa-strength-reduction.c
6642 (find_candidates_dom_walker::before_dom_children): Also allow
6645 2017-09-06 Richard Biener <rguenther@suse.de>
6647 PR tree-optimization/82108
6648 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
6649 for gap in the non-permutation SLP case.
6651 2017-09-06 Martin Jambor <mjambor@suse.cz>
6653 PR tree-optimization/82078
6654 * tree-sra.c (sort_and_splice_var_accesses): Move call to
6655 add_access_to_work_queue...
6656 (build_accesses_from_assign): ...here.
6657 (propagate_all_subaccesses): Make sure racc is the group
6658 representative, if there is one.
6660 2017-09-06 Jakub Jelinek <jakub@redhat.com>
6663 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
6666 2017-09-06 Richard Biener <rguenther@suse.de>
6668 * gimple-ssa-strength-reduction.c
6669 (find_candidates_dom_walker::before_doom_children): Use a
6670 type and not a mode check.
6672 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
6675 * config/arm/predicates.md (arm_general_adddi_operand): Create new
6677 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
6679 2017-09-05 Jeff Law <law@redhat.com>
6681 PR tree-optimization/64910
6682 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
6683 cases where we have 3 or more operands.
6685 2017-09-05 Jakub Jelinek <jakub@redhat.com>
6688 * omp-low.c (lower_omp_for): Recompute tree invariant if
6689 gimple_omp_for_initial/final is ADDR_EXPR.
6692 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
6693 into gimple val before gimplification fo the COND_EXPR.
6695 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
6697 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
6698 REGION_COPY argument.
6699 (thread_through_all_blocks): Remove unused argument to
6700 duplicate_thread_path.
6702 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6703 Alan Hayward <alan.hayward@arm.com>
6704 David Sherwood <david.sherwood@arm.com>
6706 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
6707 Take a scalar_mode rather than a machine_mode.
6708 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6709 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
6710 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6711 (aarch64_gen_adjusted_ldpstp): Likewise.
6712 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
6714 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6715 Alan Hayward <alan.hayward@arm.com>
6716 David Sherwood <david.sherwood@arm.com>
6718 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
6719 Take a scalar_int_mode instead of a machine_mode.
6720 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6721 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6722 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6723 (aarch64_simd_attr_length_rglist): Delete.
6724 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
6725 a scalar_int_mode instead of a machine_mode.
6726 (aarch64_add_offset): Likewise.
6727 (aarch64_internal_mov_immediate): Likewise
6728 (aarch64_add_constant_internal): Likewise.
6729 (aarch64_add_constant): Likewise.
6730 (aarch64_movw_imm): Likewise.
6731 (aarch64_rtx_arith_op_extract_p): Likewise.
6732 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6733 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6734 Remove assert that the mode isn't a vector.
6735 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6736 (aarch64_expand_mov_immediate): Update calls after above changes.
6737 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
6738 (aarch64_and_bitmask_imm): Check for scalar integer modes.
6739 (aarch64_move_imm): Likewise.
6740 (aarch64_can_const_movi_rtx_p): Likewise.
6741 (aarch64_strip_extend): Likewise.
6742 (aarch64_extr_rtx_p): Likewise.
6743 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
6744 a CONST_INT when the mode parameter is VOIDmode.
6745 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
6747 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6749 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
6750 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6751 (bitwise_type_for_mode): Update accordingly.
6753 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6755 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
6756 * stor-layout.c (mode_for_size_tree): Likewise.
6757 (mode_for_array): Update accordingly.
6758 (layout_decl): Likewise.
6759 (compute_record_mode): Likewise. Only set the mode once.
6761 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6763 * target.def (get_mask_mode): Change return type to opt_mode.
6765 * doc/tm.texi: Regenerate.
6766 * targhooks.h (default_get_mask_mode): Return an opt_mode.
6767 * targhooks.c (default_get_mask_mode): Likewise.
6768 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6769 * optabs-query.c (can_vec_mask_load_store_p): Update use of
6770 targetm.get_mask_mode.
6771 * tree.c (build_truth_vector_type): Likewise.
6773 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6775 * machmode.h (mode_for_vector): Return an opt_mode.
6776 * stor-layout.c (mode_for_vector): Likewise.
6777 (mode_for_int_vector): Update accordingly.
6778 (layout_type): Likewise.
6779 * config/i386/i386.c (emit_memmov): Likewise.
6780 (ix86_expand_set_or_movmem): Likewise.
6781 (ix86_expand_vector_init): Likewise.
6782 (ix86_get_mask_mode): Likewise.
6783 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
6785 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
6786 * expmed.c (extract_bit_field_1): Likewise.
6787 * expr.c (expand_expr_real_2): Likewise.
6788 * optabs-query.c (can_vec_perm_p): Likewise.
6789 (can_vec_mask_load_store_p): Likewise.
6790 * optabs.c (expand_vec_perm): Likewise.
6791 * targhooks.c (default_get_mask_mode): Likewise.
6792 * tree-vect-stmts.c (vectorizable_store): Likewise.
6793 (vectorizable_load): Likewise.
6794 (get_vectype_for_scalar_type_and_size): Likewise.
6796 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6798 * machmode.h (mode_for_int_vector): New function.
6799 * stor-layout.c (mode_for_int_vector): Likewise.
6800 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
6801 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
6802 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6803 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6804 (s390_expand_vcond): Likewise.
6806 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6808 * machmode.h (opt_machine_mode): New type.
6809 (opt_mode<T>): Allow construction from anything that can be
6811 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
6812 (mode_for_size): Return an opt_machine_mode.
6813 * stor-layout.c (mode_for_size): Likewise.
6814 (mode_for_size_tree): Update call accordingly.
6815 (bitwise_mode_for_mode): Likewise.
6816 (make_fract_type): Likewise.
6817 (make_accum_type): Likewise.
6818 * caller-save.c (replace_reg_with_saved_mem): Update call
6820 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6821 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6822 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6823 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6824 * expmed.c (extract_bit_field_1): Likewise.
6825 * reload.c (get_secondary_mem): Likewise.
6826 * varasm.c (assemble_integer): Likewise.
6827 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
6830 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6832 * machmode.h (decimal_float_mode_for_size): New function.
6833 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
6834 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
6835 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
6836 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
6837 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
6838 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
6840 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6842 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
6843 (get_builtin_sync_mode): Likewise.
6844 (expand_ifn_atomic_compare_exchange): Likewise.
6845 (expand_builtin_atomic_clear): Likewise.
6846 (expand_builtin_atomic_test_and_set): Likewise.
6847 (fold_builtin_atomic_always_lock_free): Likewise.
6848 * calls.c (compute_argument_addresses): Likewise.
6849 (emit_library_call_value_1): Likewise.
6850 (store_one_arg): Likewise.
6851 * combine.c (combine_instructions): Likewise.
6852 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
6853 * config/arm/arm.c (arm_function_value): Likewise.
6854 (aapcs_allocate_return_reg): Likewise.
6855 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
6856 * config/i386/i386.c (construct_container): Likewise.
6857 (ix86_gimplify_va_arg): Likewise.
6858 (ix86_expand_sse_cmp): Likewise.
6859 (emit_memmov): Likewise.
6860 (emit_memset): Likewise.
6861 (expand_small_movmem_or_setmem): Likewise.
6862 (ix86_expand_pextr): Likewise.
6863 (ix86_expand_pinsr): Likewise.
6864 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
6865 * config/microblaze/microblaze.c (microblaze_block_move_straight):
6867 * config/mips/mips.c (mips_function_value_1) Likewise.
6868 (mips_block_move_straight): Likewise.
6869 (mips_expand_ins_as_unaligned_store): Likewise.
6870 * config/powerpcspe/powerpcspe.c
6871 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6872 (rs6000_darwin64_record_arg_flush): Likewise.
6873 * config/rs6000/rs6000.c
6874 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6875 (rs6000_darwin64_record_arg_flush): Likewise.
6876 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
6877 (sparc_function_value_1): Likewise.
6878 * config/spu/spu.c (adjust_operand): Likewise.
6879 (spu_emit_branch_or_set): Likewise.
6880 (arith_immediate_p): Likewise.
6881 * emit-rtl.c (gen_lowpart_common): Likewise.
6882 * expr.c (expand_expr_real_1): Likewise.
6883 * function.c (assign_parm_setup_block): Likewise.
6884 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6885 * reload1.c (alter_reg): Likewise.
6886 * stor-layout.c (mode_for_vector): Likewise.
6887 (layout_type): Likewise.
6889 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6891 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
6892 (spu_convert_move): Likewise.
6893 * lower-subreg.c (resolve_simple_move): Likewise.
6895 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6898 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
6899 define_insn to a define_expand.
6900 (altivec_vsum2sws_direct): New define_insn.
6901 (altivec_vsumsws): Convert from a define_insn to a define_expand.
6903 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
6905 * config/arm/arm.c (arm_option_params_internal): Improve setting of
6908 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
6913 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
6914 realign stack if stack alignment needed is less than incoming
6917 2017-09-05 Marek Polacek <polacek@redhat.com>
6920 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
6923 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
6925 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
6927 2017-09-05 Richard Biener <rguenther@suse.de>
6929 PR tree-optimization/82084
6930 * fold-const.c (can_native_encode_string_p): Handle wide characters.
6932 2017-09-05 Richard Biener <rguenther@suse.de>
6934 PR tree-optimization/82102
6935 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
6937 2017-09-05 Martin Liska <mliska@suse.cz>
6939 PR tree-optimization/82032
6940 * tree-cfg.c (generate_range_test): New function.
6941 * tree-cfg.h (generate_range_test): Declared here.
6942 * tree-cfgcleanup.c (convert_single_case_switch): New function.
6943 (cleanup_control_expr_graph): Use it.
6944 * tree-switch-conversion.c (try_switch_expansion): Remove
6946 (emit_case_nodes): Use generate_range_test.
6948 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
6951 * config/i386/i386.md (*<btsc><mode>_mask): Add
6952 TARGET_USE_BT to insn constraint.
6953 (*btr<mode>_mask): Ditto.
6955 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
6957 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
6958 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
6960 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
6963 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
6964 TARGET_NEON and TARGET_IWMMXT.
6965 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
6966 TARGET_NEON and TARGET_IWMMXT.
6967 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
6969 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
6971 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
6972 (ix86_rewrite_tls_address): Ditto.
6973 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
6974 (ix86_rewrite_tls_address_1): Ditto.
6975 (ix86_rewrite_tls_address): Ditto.
6976 * config/i386/predicates.md (tls_address_pattern): New predicate.
6977 * config/i386/i386.md (TLS address splitter): New splitter.
6979 2017-09-04 Richard Biener <rguenther@suse.de>
6981 PR tree-optimization/82084
6982 * fold-const.h (can_native_encode_string_p): Declare.
6983 * fold-const.c (can_native_encode_string_p): Factor out from ...
6984 (native_encode_string): ... here.
6985 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
6986 vectorizing stores from constants we later cannot handle.
6988 2017-09-04 Marek Polacek <polacek@redhat.com>
6991 * doc/invoke.texi: Update -Wtautological-compare documentation.
6993 2017-09-04 Jeff Law <law@redhat.com>
6995 PR tree-optimization/64910
6996 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
6997 swap the first and last operand if the last is a constant.
6999 2017-09-04 Marek Polacek <polacek@redhat.com>
7002 * convert.c (do_narrow): When sanitizing signed integer overflows,
7003 bail out for signed types.
7004 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
7006 2017-09-04 Richard Biener <rguenther@suse.de>
7008 PR tree-optimization/82060
7009 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
7010 Move devirtualization after stmt folding and before EH/AB/noreturn
7011 cleanup to get the stmt refs canonicalized. Use a bool instead
7012 of gimple_modified_p since that doesn't work for NOPs. Schedule
7013 NOPs generated by folding for removal.
7015 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7016 Alan Hayward <alan.hayward@arm.com>
7017 David Sherwood <david.sherwood@arm.com>
7019 * coretypes.h (pad_direction): New enum.
7020 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
7021 (FUNCTION_ARG_PADDING): Likewise.
7022 * target.def (function_arg_padding): New hook.
7023 * targhooks.h (default_function_arg_padding): Declare.
7024 * targhooks.c (default_function_arg_padding): New function.
7025 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
7026 (TARGET_FUNCTION_ARG_PADDING): ...this.
7027 * doc/tm.texi: Regenerate.
7028 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
7029 instead of direction.
7030 (compute_argument_addresses): Likewise.
7031 (load_register_parameters): Likewise.
7032 (emit_library_call_value_1): Likewise.
7033 (store_one_arg): Use targetm.calls.function_arg_padding instead
7034 of FUNCTION_ARG_PADDING.
7035 (must_pass_in_stack_var_size_or_pad): Likewise.
7036 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
7037 (emit_group_store): Likewise.
7038 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
7039 instead of FUNCTION_ARG_PADDING.
7040 (emit_push_insn): Likewise, and propagate enum change throughout
7042 * function.h (direction): Delete.
7043 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
7045 * function.c (assign_parm_find_stack_rtl): Likewise.
7046 (assign_parm_setup_block_p): Likewise.
7047 (assign_parm_setup_block): Likewise.
7048 (gimplify_parameters): Likewise.
7049 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
7050 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
7052 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
7053 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7054 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
7055 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
7056 (aarch64_function_arg_padding): ...this new function.
7057 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
7058 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7059 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
7060 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7061 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
7062 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7063 (arm_pad_arg_upward): Replace with...
7064 (arm_function_arg_padding): ...this new function.
7065 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
7067 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
7068 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
7069 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7070 (ia64_hpux_function_arg_padding): Replace with...
7071 (ia64_function_arg_padding): ...this new function. Use pad_direction
7072 instead of direction. Check for TARGET_HPUX.
7073 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
7074 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7075 (iq2000_function_arg_padding): New function.
7076 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
7077 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
7078 (mips_function_arg_padding): ...this new function.
7079 (mips_pad_reg_upward): Update accordingly.
7080 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7081 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
7082 targetm.calls.function_arg_padding.
7083 (FUNCTION_ARG_PADDING): Delete.
7084 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7085 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
7086 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7087 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
7088 (nios2_block_reg_padding): Return pad_direction instead of direction.
7089 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
7090 instead of direction.
7091 (nios2_function_arg_padding): Likewise. Make static.
7092 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7093 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
7094 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
7095 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
7096 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7097 (pa_function_arg_padding): Make static. Return pad_direction instead
7099 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
7100 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7101 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
7102 instead of direction. Use targetm.calls.function_arg_padding.
7103 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
7104 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
7105 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
7106 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
7107 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
7109 (function_arg_padding): Rename to...
7110 (rs6000_function_arg_padding): ...this. Make static. Return
7111 pad_direction instead of direction.
7112 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
7113 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
7114 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7115 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
7116 instead of direction. Use targetm.calls.function_arg_padding.
7117 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
7118 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
7119 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
7120 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
7121 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7122 (function_arg_padding): Rename to...
7123 (rs6000_function_arg_padding): ...this. Make static. Return
7124 pad_direction instead of direction.
7125 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
7126 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
7127 * config/s390/s390.c (s390_function_arg_padding): New function.
7128 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7129 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
7130 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
7131 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7132 (function_arg_padding): Rename to...
7133 (sparc_function_arg_padding): ...this. Make static. Return
7134 pad_direction instead of direction.
7135 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
7136 * config/spu/spu.c (spu_function_arg_padding): New function.
7137 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7138 * system.h (FUNCTION_ARG_PADDING): Poison.
7140 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7141 Alan Hayward <alan.hayward@arm.com>
7142 David Sherwood <david.sherwood@arm.com>
7144 * target.def (modes_tieable_p): New hook.
7145 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
7146 (TARGET_MODES_TIEABLE_P): ...this.
7147 * doc/tm.texi.in: Regenerate.
7148 * hooks.h (hook_bool_mode_mode_true): Declare.
7149 * hooks.c (hook_bool_mode_mode_true): New function.
7150 * combine.c (subst): Use targetm.modes_tieable_p instead of
7152 * dse.c (find_shift_sequence): Likewise.
7153 * expmed.c (extract_low_bits): Likewise.
7154 * lower-subreg.c: Include target.h.
7155 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
7157 * rtlanal.c (rtx_cost): Likewise.
7158 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
7159 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
7160 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
7161 (TARGET_MODES_TIEABLE_P): Redefine.
7162 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
7163 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
7164 (TARGET_MODES_TIEABLE_P): Redefine.
7165 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
7166 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
7167 (arc_modes_tieable_p): New function.
7168 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
7169 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
7170 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
7171 (arm_modes_tieable_p): Make static.
7172 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
7173 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
7174 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
7175 (TARGET_MODES_TIEABLE_P): Redefine.
7176 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
7177 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
7178 (TARGET_MODES_TIEABLE_P): Redefine.
7179 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
7180 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
7181 (cr16_modes_tieable_p): New function.
7182 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
7183 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
7184 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
7185 (TRULY_NOOP_TRUNCATION): Update comment.
7186 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
7187 (TRULY_NOOP_TRUNCATION): Update comment.
7188 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
7189 (frv_modes_tieable_p): New function.
7190 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
7191 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
7192 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
7193 (TARGET_MODES_TIEABLE_P): Redefine.
7194 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
7195 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
7196 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
7197 (TARGET_MODES_TIEABLE_P): Redefine.
7198 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
7199 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
7200 (ia64_modes_tieable_p): New function.
7201 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
7202 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
7203 (iq2000_modes_tieable_p): New function.
7204 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
7205 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
7206 (lm32_modes_tieable_p): New function.
7207 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
7208 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
7209 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
7210 (TARGET_MODES_TIEABLE_P): Redefine.
7211 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
7212 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
7213 (m32r_modes_tieable_p): New function.
7214 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
7215 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
7216 (m68k_modes_tieable_p): New function.
7217 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
7218 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
7219 (mcore_modes_tieable_p): New function.
7220 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
7221 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
7223 (TARGET_MODES_TIEABLE_P): Redefine.
7224 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
7225 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
7226 * config/mips/mips.c (mips_modes_tieable_p): Make static.
7227 (TARGET_MODES_TIEABLE_P): Redefine.
7228 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
7229 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
7230 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
7231 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
7232 (mn10300_modes_tieable_p): ...this and make static.
7233 (TARGET_MODES_TIEABLE_P): Redefine.
7234 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
7235 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
7236 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
7237 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
7238 (msp430_modes_tieable_p): Make static.
7239 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
7240 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
7241 (TARGET_MODES_TIEABLE_P): Redefine.
7242 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
7243 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
7244 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
7245 (TARGET_MODES_TIEABLE_P): Redefine.
7246 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
7247 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
7248 * config/pa/pa.c (pa_modes_tieable_p): Make static.
7249 (TARGET_MODES_TIEABLE_P): Redefine.
7250 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
7251 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
7252 (pdp11_modes_tieable_p): New function.
7253 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
7254 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
7255 (rs6000_modes_tieable_p): New function.
7256 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
7257 * config/powerpcspe/powerpcspe.md: Update comment.
7258 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
7259 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
7260 (TARGET_MODES_TIEABLE_P): Redefine.
7261 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
7262 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
7263 (rl78_modes_tieable_p): New function.
7264 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
7265 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
7266 (rs6000_modes_tieable_p): New function.
7267 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
7268 * config/rs6000/rs6000.md: Update comment.
7269 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
7270 * config/rx/rx.c (rx_modes_tieable_p): New function.
7271 (TARGET_MODES_TIEABLE_P): Redefine.
7272 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
7273 * config/s390/s390.c (s390_modes_tieable_p): New function.
7274 (TARGET_MODES_TIEABLE_P): Redefine.
7275 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
7276 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
7277 (sh_modes_tieable_p): New function.
7278 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
7279 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
7280 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
7281 (sparc_modes_tieable_p): Make static.
7282 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
7283 * config/spu/spu.c (spu_modes_tieable_p): New function.
7284 (TARGET_MODES_TIEABLE_P): Redefine.
7285 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
7286 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
7287 (TARGET_MODES_TIEABLE_P): Redefine.
7288 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
7289 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
7290 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
7291 * config/v850/v850.c (v850_modes_tieable_p): New function.
7292 (TARGET_MODES_TIEABLE_P): Redefine.
7293 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
7294 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
7295 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
7296 (visium_modes_tieable_p): New function.
7297 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
7298 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
7299 (xtensa_modes_tieable_p): New function.
7300 * system.h (MODES_TIEABLE_P): Poison.
7302 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7303 Alan Hayward <alan.hayward@arm.com>
7304 David Sherwood <david.sherwood@arm.com>
7306 * target.def (hard_regno_mode_ok): New hook.
7307 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
7308 (TARGET_HARD_REGNO_MODE_OK): ...this.
7309 * doc/tm.texi.in: Regenerate.
7310 * hooks.h (hook_bool_uint_mode_true): Declare.
7311 * hooks.c (hook_bool_uint_mode_true): New function.
7312 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
7314 * genpreds.c (write_insn_preds_c): Add an include of target.h.
7315 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
7316 instead of HARD_REGNO_MODE_OK.
7317 * caller-save.c: Include target.h.
7318 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
7320 * combine.c (can_combine_p): Likewise.
7321 (combinable_i3pat): Likewise.
7322 (can_change_dest_mode): Likewise.
7323 * expr.c (init_expr_target): Likewise.
7324 (convert_move): Likewise.
7325 (convert_modes): Likewise.
7326 * ira.c (setup_prohibited_class_mode_regs): Likewise.
7327 (setup_prohibited_mode_move_regs): Likewise.
7328 * ira.h (target_ira): Likewise.
7329 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7330 * lra-constraints.c (process_alt_operands): Likewise.
7331 (split_reg): Likewise.
7332 * recog.c (peep2_find_free_register): Likewise.
7333 * ree.c (combine_reaching_defs): Likewise.
7334 * regcprop.c (maybe_mode_change): Likewise.
7335 * reginfo.c (init_reg_sets_1): Likewise.
7336 (choose_hard_reg_mode): Likewise.
7337 (simplifiable_subregs): Likewise.
7338 * regrename.c (check_new_reg_p): Likewise.
7339 * reload.c (find_valid_class): Likewise.
7340 (find_valid_class_1): Likewise.
7341 (reload_inner_reg_of_subreg): Likewise.
7342 (push_reload): Likewise.
7343 (combine_reloads): Likewise.
7344 (find_dummy_reload): Likewise.
7345 (find_reloads): Likewise.
7346 * reload1.c (find_reg): Likewise.
7347 (set_reload_reg): Likewise.
7348 (allocate_reload_reg): Likewise.
7349 (choose_reload_regs): Likewise.
7350 (reload_adjust_reg_for_temp): Likewise.
7351 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
7352 (simplify_subreg_regno): Likewise.
7353 * sel-sched.c (init_regs_for_mode): Likewise.
7354 * varasm.c (make_decl_rtl): Likewise.
7355 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
7356 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
7358 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
7359 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
7360 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7361 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
7362 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
7363 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7364 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
7365 (arc_mode_class): Delete.
7366 (HARD_REGNO_MODE_OK): Delete.
7367 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7368 (arc_hard_regno_mode_ok): Rename old array to...
7369 (arc_hard_regno_mode_ok_modes): ...this.
7370 (arc_conditional_register_usage): Update accordingly.
7371 (arc_mode_class): Make static.
7372 (arc_hard_regno_mode_ok): New function.
7373 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
7374 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
7375 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7376 (arm_hard_regno_mode_ok): Make static.
7377 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
7379 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
7380 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
7381 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
7383 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7384 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
7385 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
7386 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
7387 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
7388 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7389 * config/bfin/predicates.md (valid_reg_operand): Use
7390 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
7391 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
7392 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
7393 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7394 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
7395 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
7396 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7397 (cr16_hard_regno_mode_ok): Make static and return a bool.
7398 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
7399 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7400 (cris_hard_regno_mode_ok): New function.
7401 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
7402 (epiphany_mode_class): Delete.
7403 (HARD_REGNO_MODE_OK): Delete.
7404 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
7405 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7406 (hard_regno_mode_ok): Rename to...
7407 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
7408 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
7409 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
7411 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
7412 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
7413 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7414 (frv_hard_regno_mode_ok): Make static and return a bool.
7415 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
7417 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
7418 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
7419 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
7420 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
7422 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7423 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
7424 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
7425 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
7427 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7428 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
7429 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7430 (ia64_hard_regno_mode_ok): New function.
7431 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
7432 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7433 (iq2000_hard_regno_mode_ok): New function.
7434 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
7435 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7436 (lm32_hard_regno_mode_ok): New function.
7437 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
7438 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
7439 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
7440 instead of HARD_REGNO_MODE_OK.
7441 (m32c_hard_regno_ok): Rename to...
7442 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
7443 (m32c_cannot_change_mode_class): Update accordingly.
7444 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7445 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
7446 (m32r_mode_class): Delete.
7447 (HARD_REGNO_MODE_OK): Delete.
7448 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7449 (m32r_hard_regno_mode_ok): Rename to...
7450 (m32r_hard_regno_modes): ...this.
7451 (m32r_mode_class): Make static.
7452 (m32r_hard_regno_mode_ok): New function.
7453 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
7454 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
7455 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7456 (m68k_hard_regno_mode_ok): Make static.
7457 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
7458 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7459 (mcore_hard_regno_mode_ok): New function.
7460 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
7461 (HARD_REGNO_MODE_OK): Delete.
7462 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
7464 (microblaze_hard_regno_mode_ok_p): ...this and make static.
7465 (microblaze_hard_regno_mode_ok): New function.
7466 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7467 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
7468 (mips_hard_regno_mode_ok): Delete.
7469 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
7470 (mips_hard_regno_mode_ok_p): ...this and make static.
7471 (mips_hard_regno_mode_ok_p): Rename to...
7472 (mips_hard_regno_mode_ok_uncached): ...this.
7473 (mips_hard_regno_mode_ok): New function.
7474 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
7475 of HARD_REGNO_MODE_OK.
7476 (mips_option_override): Update after above name changes.
7477 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7478 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
7479 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
7480 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
7481 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
7482 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7483 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
7484 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
7485 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
7486 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7487 (msp430_hard_regno_mode_ok): Make static and return a bool.
7488 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
7489 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
7490 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
7492 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7493 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
7494 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
7495 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
7496 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
7497 (PA_HARD_REGNO_MODE_OK): ...this
7498 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
7499 (PA_HARD_REGNO_MODE_OK): ...this.
7500 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7501 (pa_hard_regno_mode_ok): New function.
7502 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
7503 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7504 (pdp11_hard_regno_mode_ok): New function.
7505 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
7506 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
7508 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
7510 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7511 (rs6000_hard_regno_mode_ok): Rename to...
7512 (rs6000_hard_regno_mode_ok_uncached): ...this.
7513 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7514 (rs6000_hard_regno_mode_ok): New function.
7515 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
7516 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
7517 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
7518 (riscv_hard_regno_mode_ok): ...this and make static.
7519 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7520 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
7521 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
7522 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7523 (rl78_hard_regno_mode_ok): Make static and return bool.
7524 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
7525 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
7527 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
7528 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7529 (rs6000_hard_regno_mode_ok): Rename to...
7530 (rs6000_hard_regno_mode_ok_uncached): ...this.
7531 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7532 (rs6000_hard_regno_mode_ok): New function.
7533 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
7534 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
7535 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7536 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
7537 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
7538 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
7539 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7540 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
7541 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
7542 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7543 (sh_hard_regno_mode_ok): Make static.
7544 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
7545 instead of HARD_REGNO_MODE_OK.
7546 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
7547 (sparc_mode_class): Delete.
7548 (HARD_REGNO_MODE_OK): Delete.
7549 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7550 (hard_regno_mode_classes): Make static.
7551 (sparc_mode_class): Likewise.
7552 (sparc_hard_regno_mode_ok): New function.
7553 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
7554 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
7555 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
7557 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7558 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
7559 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
7560 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
7561 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
7562 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7563 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
7564 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
7565 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7566 (visium_hard_regno_mode_ok): New function.
7567 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
7568 instead of HARD_REGNO_MODE_OK.
7569 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
7570 (HARD_REGNO_MODE_OK): Delete.
7571 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
7572 (xtensa_hard_regno_mode_ok_p): ...this and make static.
7573 (xtensa_option_override): Update accordingly.
7574 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7575 (xtensa_hard_regno_mode_ok): New function.
7576 * system.h (HARD_REGNO_MODE_OK): Poison.
7578 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7579 Alan Hayward <alan.hayward@arm.com>
7580 David Sherwood <david.sherwood@arm.com>
7582 * target.def (hard_regno_call_part_clobbered): New hook.
7583 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
7584 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
7585 * doc/tm.texi: Regenerate.
7586 * hooks.h (hook_bool_uint_mode_false): Declare.
7587 * hooks.c (hook_bool_uint_mode_false): New function.
7588 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7589 * cselib.c (cselib_process_insn): Use
7590 targetm.hard_regno_call_part_clobbered instead of
7591 HARD_REGNO_CALL_PART_CLOBBERED.
7592 * ira-conflicts.c (ira_build_conflicts): Likewise.
7593 * ira-costs.c (ira_tune_allocno_costs): Likewise.
7594 * lra-constraints.c (need_for_call_save_p): Likewise.
7595 * lra-lives.c: Include target.h.
7596 (check_pseudos_live_through_calls): Use
7597 targetm.hard_regno_call_part_clobbered instead of
7598 HARD_REGNO_CALL_PART_CLOBBERED.
7599 * regcprop.c: Include target.h.
7600 (copyprop_hardreg_forward_1): Use
7601 targetm.hard_regno_call_part_clobbered instead of
7602 HARD_REGNO_CALL_PART_CLOBBERED.
7603 * reginfo.c (choose_hard_reg_mode): Likewise.
7604 * regrename.c (check_new_reg_p): Likewise.
7605 * reload.c (find_equiv_reg): Likewise.
7606 * reload1.c (emit_reload_insns): Likewise.
7607 * sched-deps.c (deps_analyze_insn): Likewise.
7608 * sel-sched.c (init_regs_for_mode): Likewise.
7609 (mark_unavailable_hard_regs): Likewise.
7610 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
7611 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7612 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
7614 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7615 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7616 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
7618 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
7620 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7621 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7622 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
7624 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7625 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7626 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
7628 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7629 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
7631 * config/powerpcspe/powerpcspe.c
7632 (rs6000_hard_regno_call_part_clobbered): New function.
7633 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7634 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7635 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
7637 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7638 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7639 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
7641 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7642 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7643 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
7645 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7646 Alan Hayward <alan.hayward@arm.com>
7647 David Sherwood <david.sherwood@arm.com>
7649 * rtl.h (subreg_memory_offset): Declare.
7650 * emit-rtl.c (subreg_memory_offset): New function.
7651 * expmed.c (store_bit_field_1): Use it.
7652 * expr.c (undefined_operand_subword_p): Likewise.
7653 * simplify-rtx.c (simplify_subreg): Likewise.
7655 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
7657 PR rtl-optimization/57448
7660 * optabs.c (expand_atomic_load): Place compiler memory barriers if
7661 using atomic_load pattern.
7662 (expand_atomic_store): Likewise.
7664 2017-09-04 Jakub Jelinek <jakub@redhat.com>
7667 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
7668 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
7669 handling. Use replace_call_with_value with NULL instead of
7670 gsi_replace, unlink_stmt_vdef and release_defs.
7672 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
7675 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
7677 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7680 * rtl.h (emit_library_call_value_1): Declare.
7681 (emit_library_call): Replace declaration with a series of overloads.
7682 Remove the parameter count argument.
7683 (emit_library_call_value): Likewise.
7684 * calls.c (emit_library_call_value_1): Make global. Replace varargs
7685 with an "rtx_mode_t *".
7686 (emit_library_call_value): Delete.
7687 (emit_library_call): Likewise.
7688 * asan.c (asan_emit_stack_protection): Update calls accordingly.
7689 (asan_emit_allocas_unpoison): Likewise.
7690 * builtins.c (expand_builtin_powi): Likewise.
7691 (expand_asan_emit_allocas_unpoison): Likewise.
7692 * cfgexpand.c (expand_main_function): Likewise.
7693 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
7694 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
7695 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
7696 * config/arm/arm.c (arm_trampoline_init): Likewise.
7697 (arm_call_tls_get_addr): Likewise.
7698 (arm_expand_divmod_libfunc): Likewise.
7699 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
7700 (smulsi3_highpart): Likewise.
7701 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
7702 (c6x_expand_compare): Likewise.
7703 (c6x_expand_movmem): Likewise.
7704 * config/frv/frv.c (frv_trampoline_init): Likewise.
7705 * config/i386/i386.c (ix86_trampoline_init): Likewise.
7706 (ix86_expand_divmod_libfunc): Likewise.
7707 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
7708 (ia64_expand_compare): Likewise.
7709 (ia64_profile_hook): Likewise.
7710 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
7711 (nonlocal_goto): Likewise.
7712 (restore_stack_nonlocal): Likewise.
7713 * config/m32r/m32r.c (block_move_call): Likewise.
7714 (m32r_trampoline_init): Likewise.
7715 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
7716 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
7717 (m68k_call_m68k_read_tp): Likewise.
7718 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
7719 (microblaze_expand_divide): Likewise.
7720 * config/mips/mips.h (mips_args): Likewise.
7721 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
7722 (MIPS_ICACHE_SYNC): Likewise.
7723 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
7724 (nios2_trampoline_init): Likewise.
7725 * config/pa/pa.c (hppa_tls_call): Likewise.
7726 (pa_trampoline_init): Likewise.
7727 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
7728 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
7729 (expand_strn_compare): Likewise.
7730 (rs6000_generate_compare): Likewise.
7731 (rs6000_expand_float128_convert): Likewise.
7732 (output_profile_hook): Likewise.
7733 (rs6000_trampoline_init): Likewise.
7734 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
7735 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
7736 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
7737 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
7738 (rs6000_generate_compare): Likewise.
7739 (rs6000_expand_float128_convert): Likewise.
7740 (output_profile_hook): Likewise.
7741 (rs6000_trampoline_init): Likewise.
7742 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
7743 * config/sh/sh.c (sh_trampoline_init): Likewise.
7744 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
7745 (sparc_emit_float_lib_cmp): Likewise.
7746 (sparc32_initialize_trampoline): Likewise.
7747 (sparc64_initialize_trampoline): Likewise.
7748 (sparc_profile_hook): Likewise.
7749 * config/spu/spu.c (ea_load_store): Likewise.
7750 * config/spu/spu.md (floatunssidf2): Likewise.
7751 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
7752 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
7753 * config/visium/visium.c (expand_block_move_4): Likewise.
7754 (expand_block_move_2): Likewise.
7755 (expand_block_move_1): Likewise.
7756 (expand_block_set_4): Likewise.
7757 (expand_block_set_2): Likewise.
7758 (expand_block_set_1): Likewise.
7759 (visium_trampoline_init): Likewise.
7760 (visium_profile_hook): Likewise.
7761 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
7762 (xtensa_setup_frame_addresses): Likewise.
7763 (xtensa_trampoline_init): Likewise.
7764 * except.c (sjlj_emit_function_enter): Likewise.
7765 (sjlj_emit_function_exit): Likewise.
7766 * explow.c (allocate_dynamic_stack_space): Likewise.
7767 (probe_stack_range): Likewise.
7768 * expr.c (convert_mode_scalar): Likewise.
7769 * optabs.c (expand_binop): Likewise.
7770 (expand_twoval_binop_libfunc): Likewise.
7771 (expand_unop): Likewise.
7772 (prepare_cmp_insn): Likewise.
7773 (prepare_float_lib_cmp): Likewise.
7774 (expand_float): Likewise.
7775 (expand_fix): Likewise.
7776 (expand_fixed_convert): Likewise.
7777 (maybe_emit_sync_lock_test_and_set): Likewise.
7778 (expand_atomic_compare_and_swap): Likewise.
7779 (expand_mem_thread_fence): Likewise.
7780 (expand_atomic_fetch_op): Likewise.
7782 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
7784 * doc/generic.texi (OpenACC): Adjust URL.
7785 * doc/invoke.texi (C Dialect Options): Ditto.
7787 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
7789 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
7790 predicate for operand 1. Add (m,<S>) constraint.
7791 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
7792 Prevent memory operand 1 with register operand 2.
7794 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
7796 PR rtl-optimization/82024
7797 * combine.c (try_combine): If the combination result is a PARALLEL,
7798 and we only need to retain the SET in there that would be placed
7799 at I2, check that we can place that at I3 instead, before doing so.
7801 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7804 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
7806 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
7807 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
7810 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
7811 Jeff Law <law@redhat.com>
7813 * varasm.c (bss_initializer_p): Do not put constants into .bss
7814 (categorize_decl_for_section): Handle bss_initializer_p returning
7815 false when DECL_INITIAL is NULL.
7817 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7820 * config/s390/s390.c (s390_can_inline_p): New function.
7822 2017-09-01 Jeff Law <law@redhat.com>
7824 PR tree-optimization/82052
7825 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
7826 Always initialize the returned slot after a hash table miss
7827 when INSERT is true.
7829 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
7831 * config/s390/s390.md (mem_signal_fence): Remove.
7832 * doc/md.texi (mem_signal_fence): Remove.
7833 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
7835 * target-insns.def (mem_signal_fence): Remove.
7837 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7840 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
7843 * asan.c (create_odr_indicator): Strip name encoding from assembler
7844 name before appending it after __odr_asan_.
7846 2017-09-01 Martin Liska <mliska@suse.cz>
7848 PR tree-optimization/82059
7849 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
7850 frequency only when an edge is redirected.
7852 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7854 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
7855 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
7856 (arc_conditional_register_usage): Remove ARC600 lp_count
7858 (arc_file_start): Emit Tag_ARC_CPU_variation.
7859 (arc_can_use_doloop_p): New conditions to use ZOLs.
7860 (hwloop_fail): New function.
7861 (hwloop_optimize): Likewise.
7862 (hwloop_pattern_reg): Likewise.
7863 (arc_doloop_hooks): New struct, to be used with reorg_loops.
7864 (arc_reorg_loops): New function, calls reorg_loops.
7865 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
7866 (arc600_corereg_hazard): Remove ZOL checking, case handled by
7868 (arc_loop_hazard): Remove function, functionality moved into
7870 (arc_hazard): Remove arc_loop_hazard call.
7871 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
7872 into hwloop_optimize.
7873 (arc_label_align): Remove ZOL handling.
7874 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
7875 * config/arc/arc.md (doloop_begin): Remove pattern.
7876 (doloop_begin_i): Likewise.
7877 (doloop_end_i): Likewise.
7878 (doloop_fallback): Likewise.
7879 (doloop_fallback_m): Likewise.
7880 (doloop_end): Reimplement expand.
7881 (arc_lp): New pattern for LP instruction.
7882 (loop_end): New pattern.
7883 (loop_fail): Likewise.
7884 (decrement_and_branch_until_zero): Likewise.
7885 * config/arc/arc.opt (mlpc-width): New option.
7886 * doc/invoke.texi (mlpc-width): Document option.
7888 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7890 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
7891 (arc_ccfsm_advance): Fix checking for delay slots.
7892 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
7894 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7896 * config/arc/arc.md (movqi_insn): Add stores to save constant long
7898 (movhi_insn): Update store instruction constraint which are saving
7899 6-bit short immediates.
7900 (movsi_insn): Consider also short scaled load operations.
7901 (zero_extendhisi2_i): Use Usd constraint instead of T.
7902 (extendhisi2_i): Add q constraint.
7903 (arc_clzsi2): Add type and length attributes.
7904 (arc_ctzsi2): Likewise.
7905 * config/arc/constraints.md (Usc): Update constraint, the
7906 assembler can parse two relocations for a single instruction.
7908 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7910 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
7911 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
7913 2017-08-31 Olivier Hainque <hainque@adacore.com>
7915 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
7916 match as powerpc-wrs-vxworks*.
7918 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
7920 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
7921 register constraint for by-element operand.
7922 (aarch64_mls_elt_merge<mode>): Likewise.
7924 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7926 * config/arc/arc.c (arc_can_follow_jump): Check for short
7929 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7931 * config.gcc: Use g.opt for arc.
7932 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
7933 functionality moved to ...
7934 (legitimate_scaled_address_p): New function, ...here.
7935 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
7936 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
7937 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
7939 (arc_override_options): Handle G option.
7940 (arc_output_pic_addr_const): Correct function definition.
7941 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
7942 (arc_decl_anon_ns_mem_p): Delete.
7943 (arc_in_small_data_p): Overhaul this function to take into
7944 consideration the value given via G option.
7945 (arc_rewrite_small_data_1): Renamed and corrected old
7946 arc_rewrite_small_data function.
7947 (arc_rewrite_small_data): New function.
7948 (small_data_pattern): Don't use pic_offset_table_rtx.
7949 * config/arc/arc.h (CC1_SPEC): Recognize G option.
7950 * config/arc/simdext.md (movmisalignv2hi): Use
7951 prepare_move_operands function.
7953 (movmisalign*): Likewise.
7954 * doc/invoke.texi (ARC options): Document -G option.
7956 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7958 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
7960 * config/arc/arc.c (arc_print_operand): Output scalled address for
7961 sdata whenever is possible.
7962 (arc_in_small_data_p): Allow sdata for 64bit datum when double
7963 load/stores are available.
7964 (compact_sda_memory_operand): Check for the alignment required by
7965 code density instructions.
7966 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
7968 * config/arc/constraints.md (Usd): Update constraint.
7969 (Us0): New constraint.
7970 (Usc): Update constraint.
7972 2017-08-31 Richard Biener <rguenther@suse.de>
7975 * dwarf2out.c (dwarf2out_early_global_decl): Process each
7978 2017-08-31 Tamar Christina <tamar.christina@arm.com>
7980 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
7981 Resize type_signature.
7983 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7984 Alan Hayward <alan.hayward@arm.com>
7985 David Sherwood <david.sherwood@arm.com>
7987 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
7988 subregs whose inner modes can be stored in GPRs.
7989 (aarch64_classify_index): Likewise.
7991 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7992 Alan Hayward <alan.hayward@arm.com>
7993 David Sherwood <david.sherwood@arm.com>
7995 * config/aarch64/iterators.md (V_cmp_result): Rename to...
7996 (V_INT_EQUIV): ...this.
7997 (v_cmp_result): Rename to...
7998 (v_int_equiv): ...this.
7999 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
8000 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
8001 (copysign<mode>3): Likewise.
8002 (aarch64_simd_bsl<mode>_internal): Likewise.
8003 (aarch64_simd_bsl<mode>): Likewise.
8004 (vec_cmp<mode><mode>): Likewise.
8005 (vcond<mode><mode>): Likewise.
8006 (vcond<v_cmp_mixed><mode>): Likewise.
8007 (vcondu<mode><v_cmp_mixed>): Likewise.
8008 (aarch64_cm<optab><mode>): Likewise.
8009 (aarch64_cmtst<mode>): Likewise.
8010 (aarch64_fac<optab><mode>): Likewise.
8011 (vec_perm_const<mode>): Likewise.
8012 (vcond_mask_<mode><v_cmp_result>): Rename to...
8013 (vcond_mask_<mode><v_int_equiv>): ...this.
8014 (vec_cmp<mode><v_cmp_result>): Rename to...
8015 (vec_cmp<mode><v_int_equiv>): ...this.
8017 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8018 Alan Hayward <alan.hayward@arm.com>
8019 David Sherwood <david.sherwood@arm.com>
8021 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
8023 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
8024 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
8025 (UNSPEC_LD4_DREG): New unspecs.
8026 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
8027 (aarch64_ld2<mode>_dreg_be): Replace with...
8028 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
8030 (aarch64_ld3<mode>_dreg_le)
8031 (aarch64_ld3<mode>_dreg_be): Replace with...
8032 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
8034 (aarch64_ld4<mode>_dreg_le)
8035 (aarch64_ld4<mode>_dreg_be): Replace with...
8036 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
8039 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8041 PR tree-optimization/81987
8042 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
8043 insert an initializer in a location not dominated by the stride
8046 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8048 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
8049 on the entire header of the finally block in the fallthru case.
8051 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8053 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
8055 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
8057 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
8058 rs6000_emit_move_from_cr and call renamed function.
8059 (rs6000_emit_prologue): Call renamed functions.
8060 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
8061 movesi_from_cr, remove volatile CRs.
8063 2017-08-30 Jon Beniston <jon@beniston.com>
8064 Richard Biener <rguenther@suse.de>
8066 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
8067 of VECTOR_MODE_P check.
8068 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
8069 element vector types.
8071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8073 * df.h (df_read_modify_subreg_p): Remove in favor of...
8074 * rtl.h (read_modify_subreg_p): ...this new function. Take a
8075 const_rtx instead of an rtx.
8076 * cprop.c (local_cprop_find_used_regs): Update accordingly.
8077 * df-problems.c (df_word_lr_mark_ref): Likewise.
8078 * ira-lives.c (mark_pseudo_reg_live): Likewise.
8079 (mark_pseudo_reg_dead): Likewise.
8080 (mark_ref_dead): Likewise.
8081 * reginfo.c (init_subregs_of_mode): Likewise.
8082 * sched-deps.c (sched_analyze_1): Likewise.
8083 * df-scan.c (df_def_record_1): Likewise.
8084 (df_uses_record): Likewise.
8085 (df_read_modify_subreg_p): Remove in favor of...
8086 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
8087 const_rtx instead of an rtx.
8089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8090 Alan Hayward <alan.hayward@arm.com>
8091 David Sherwood <david.sherwood@arm.com>
8093 * rtl.h (partial_subreg_p): New function.
8094 * caller-save.c (save_call_clobbered_regs): Use it.
8095 * calls.c (expand_call): Likewise.
8096 * combine.c (combinable_i3pat): Likewise.
8097 (simplify_set): Likewise.
8098 (make_extraction): Likewise.
8099 (make_compound_operation_int): Likewise.
8100 (gen_lowpart_or_truncate): Likewise.
8101 (force_to_mode): Likewise.
8102 (make_field_assignment): Likewise.
8103 (reg_truncated_to_mode): Likewise.
8104 (record_truncated_value): Likewise.
8105 (move_deaths): Likewise.
8106 * cse.c (record_jump_cond): Likewise.
8107 (cse_insn): Likewise.
8108 * cselib.c (cselib_lookup_1): Likewise.
8109 * expmed.c (extract_bit_field_using_extv): Likewise.
8110 * function.c (assign_parm_setup_reg): Likewise.
8111 * ifcvt.c (noce_convert_multiple_sets): Likewise.
8112 * ira-build.c (create_insn_allocnos): Likewise.
8113 * lra-coalesce.c (merge_pseudos): Likewise.
8114 * lra-constraints.c (match_reload): Likewise.
8115 (simplify_operand_subreg): Likewise.
8116 (curr_insn_transform): Likewise.
8117 * lra-lives.c (process_bb_lives): Likewise.
8118 * lra.c (new_insn_reg): Likewise.
8119 (lra_substitute_pseudo): Likewise.
8120 * regcprop.c (mode_change_ok): Likewise.
8121 (maybe_mode_change): Likewise.
8122 (copyprop_hardreg_forward_1): Likewise.
8123 * reload.c (push_reload): Likewise.
8124 (find_reloads): Likewise.
8125 (find_reloads_subreg_address): Likewise.
8126 * reload1.c (alter_reg): Likewise.
8127 (eliminate_regs_1): Likewise.
8128 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8130 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
8132 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
8133 back to if statements, including unpack.
8135 2017-08-30 Martin Liska <mliska@suse.cz>
8138 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
8140 (func_checker::compare_asm_inputs_outputs): ... this function.
8141 (func_checker::compare_gimple_asm): Use the function to compare
8142 also ASM constrains.
8143 * ipa-icf-gimple.h: Rename the function.
8145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8146 Alan Hayward <alan.hayward@arm.com>
8147 David Sherwood <david.sherwood@arm.com>
8149 * coretypes.h (complex_mode): New type.
8150 * gdbhooks.py (build_pretty_printer): Handle it.
8151 * machmode.h (complex_mode): New class.
8152 (complex_mode::includes_p): New function.
8153 (is_complex_int_mode): Likewise.
8154 (is_complex_float_mode): Likewise.
8155 * genmodes.c (get_mode_class): Handle complex mode classes.
8156 * function.c (expand_function_end): Use is_complex_int_mode.
8158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8159 Alan Hayward <alan.hayward@arm.com>
8160 David Sherwood <david.sherwood@arm.com>
8162 * coretypes.h (scalar_mode_pod): New typedef.
8163 * gdbhooks.py (build_pretty_printer): Handle it.
8164 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
8165 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
8166 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
8167 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
8170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8171 Alan Hayward <alan.hayward@arm.com>
8172 David Sherwood <david.sherwood@arm.com>
8174 * machmode.h (mode_for_vector): Take a scalar_mode instead
8176 * stor-layout.c (mode_for_vector): Likewise.
8177 * explow.c (promote_mode): Use as_a <scalar_mode>.
8178 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
8180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8181 Alan Hayward <alan.hayward@arm.com>
8182 David Sherwood <david.sherwood@arm.com>
8184 * target.def (preferred_simd_mode): Take a scalar_mode
8185 instead of a machine_mode.
8186 * targhooks.h (default_preferred_simd_mode): Likewise.
8187 * targhooks.c (default_preferred_simd_mode): Likewise.
8188 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
8189 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
8190 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
8191 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
8192 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
8193 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
8194 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
8195 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
8197 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
8198 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
8199 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
8200 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
8201 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
8202 * doc/tm.texi: Regenerate.
8203 * optabs-query.c (can_vec_mask_load_store_p): Return false for
8206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8207 Alan Hayward <alan.hayward@arm.com>
8208 David Sherwood <david.sherwood@arm.com>
8210 * target.def (scalar_mode_supported_p): Take a scalar_mode
8211 instead of a machine_mode.
8212 * targhooks.h (default_scalar_mode_supported_p): Likewise.
8213 * targhooks.c (default_scalar_mode_supported_p): Likewise.
8214 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
8215 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8216 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
8217 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
8218 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
8219 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
8220 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
8221 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
8222 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
8223 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
8224 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
8225 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
8227 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
8228 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
8229 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8230 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8231 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
8233 * doc/tm.texi: Regenerate.
8235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8236 Alan Hayward <alan.hayward@arm.com>
8237 David Sherwood <david.sherwood@arm.com>
8239 * coretypes.h (opt_scalar_mode): New typedef.
8240 * gdbhooks.py (build_pretty_printers): Handle it.
8241 * machmode.h (mode_iterator::get_2xwider): Add overload for
8243 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
8245 * expr.c (convert_mode_scalar): Likewise.
8246 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8247 * optabs.c (expand_float): Likewise.
8248 (expand_fix): Likewise.
8249 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8252 Alan Hayward <alan.hayward@arm.com>
8253 David Sherwood <david.sherwood@arm.com>
8255 * optabs.c (expand_float): Explicitly check for scalars before
8256 using a branching expansion.
8257 (expand_fix): Likewise.
8259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8260 Alan Hayward <alan.hayward@arm.com>
8261 David Sherwood <david.sherwood@arm.com>
8263 * expr.c (convert_mode): Split scalar handling out into...
8264 (convert_mode_scalar): ...this new function. Treat the modes
8267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8268 Alan Hayward <alan.hayward@arm.com>
8269 David Sherwood <david.sherwood@arm.com>
8271 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
8273 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
8275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8276 Alan Hayward <alan.hayward@arm.com>
8277 David Sherwood <david.sherwood@arm.com>
8279 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
8280 rather than a machine_mode.
8281 (fixed_from_string): Likewise.
8282 (fixed_convert): Likewise.
8283 (fixed_convert_from_int): Likewise.
8284 (fixed_convert_from_real): Likewise.
8285 (real_convert_from_fixed): Likewise.
8286 * fixed-value.c (fixed_from_double_int): Likewise.
8287 (fixed_from_string): Likewise.
8288 (fixed_convert): Likewise.
8289 (fixed_convert_from_int): Likewise.
8290 (fixed_convert_from_real): Likewise.
8291 (real_convert_from_fixed): Likewise.
8292 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
8294 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8295 Alan Hayward <alan.hayward@arm.com>
8296 David Sherwood <david.sherwood@arm.com>
8298 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
8299 of separate mode class checks. Do not allow vector modes here.
8300 (immed_wide_int_const): Use as_a <scalar_mode>.
8301 * explow.c (trunc_int_for_mode): Likewise.
8302 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
8303 (wi::shwi): Likewise.
8304 (wi::min_value): Likewise.
8305 (wi::max_value): Likewise.
8306 * dwarf2out.c (loc_descriptor): Likewise.
8307 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
8310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8311 Alan Hayward <alan.hayward@arm.com>
8312 David Sherwood <david.sherwood@arm.com>
8314 * tree.h (SCALAR_TYPE_MODE): New macro.
8315 * expr.c (expand_expr_addr_expr_1): Use it.
8316 (expand_expr_real_2): Likewise.
8317 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
8318 (fold_convert_const_fixed_from_int): Likewise.
8319 (fold_convert_const_fixed_from_real): Likewise.
8320 (native_encode_fixed): Likewise
8321 (native_encode_complex): Likewise
8322 (native_encode_vector): Likewise.
8323 (native_interpret_fixed): Likewise.
8324 (native_interpret_real): Likewise.
8325 (native_interpret_complex): Likewise.
8326 (native_interpret_vector): Likewise.
8327 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
8328 (simd_clone_adjust_argument_types): Likewise.
8329 (simd_clone_init_simd_arrays): Likewise.
8330 (simd_clone_adjust): Likewise.
8331 * stor-layout.c (layout_type): Likewise.
8332 * tree.c (build_minus_one_cst): Likewise.
8333 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
8334 * tree-inline.c (estimate_move_cost): Likewise.
8335 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
8336 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
8337 (vectorizable_reduction): Likewise.
8338 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
8339 (vect_recog_mixed_size_cond_pattern): Likewise.
8340 (check_bool_pattern): Likewise.
8341 (adjust_bool_pattern): Likewise.
8342 (search_type_for_mask_1): Likewise.
8343 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
8344 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8345 (vectorizable_load): Likewise.
8346 (vectorizable_store): Likewise.
8347 * ubsan.c (ubsan_encode_value): Likewise.
8348 * varasm.c (output_constant): Likewise.
8350 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8351 Alan Hayward <alan.hayward@arm.com>
8352 David Sherwood <david.sherwood@arm.com>
8354 * coretypes.h (scalar_mode): New class.
8355 * machmode.h (scalar_mode): Likewise.
8356 (scalar_mode::includes_p): New function.
8357 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
8358 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
8359 * genmodes.c (get_mode_class): Handle remaining scalar modes.
8360 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
8361 * expmed.c (store_bit_field_1): Likewise.
8362 (extract_bit_field_1): Likewise.
8363 * expr.c (write_complex_part): Likewise.
8364 (read_complex_part): Likewise.
8365 (emit_move_complex_push): Likewise.
8366 (expand_expr_real_2): Likewise.
8367 * function.c (assign_parm_setup_reg): Likewise.
8368 (assign_parms_unsplit_complex): Likewise.
8369 * optabs.c (expand_binop): Likewise.
8370 * rtlanal.c (subreg_get_info): Likewise.
8371 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8372 * varasm.c (output_constant_pool_2): Likewise.
8374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8375 Alan Hayward <alan.hayward@arm.com>
8376 David Sherwood <david.sherwood@arm.com>
8378 * expmed.c (extract_high_half): Use scalar_int_mode and remove
8380 (expmed_mult_highpart_optab): Likewise.
8381 (expmed_mult_highpart): Likewise.
8383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8384 Alan Hayward <alan.hayward@arm.com>
8385 David Sherwood <david.sherwood@arm.com>
8387 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
8388 instead of a machine_mode.
8389 (builtin_memset_read_str): Likewise.
8390 * builtins.c (c_readstr): Likewise.
8391 (builtin_memcpy_read_str): Likewise.
8392 (builtin_strncpy_read_str): Likewise.
8393 (builtin_memset_read_str): Likewise.
8394 (builtin_memset_gen_str): Likewise.
8395 (expand_builtin_signbit): Use scalar_int_mode for local variables.
8396 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
8397 instead of a machine_mode.
8398 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
8400 (make_extraction): Likewise.
8401 (try_widen_shift_mode): Take and return scalar_int_modes instead
8403 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
8404 a scalar_int_mode instead of a machine_mode.
8405 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
8406 (avr_addr_space_pointer_mode): Likewise.
8407 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
8408 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
8409 (msp430_unwind_word_mode): Likewise.
8410 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
8411 (spu_addr_space_pointer_mode): Likewise.
8412 (spu_addr_space_address_mode): Likewise.
8413 (spu_libgcc_cmp_return_mode): Likewise.
8414 (spu_libgcc_shift_count_mode): Likewise.
8415 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
8416 (rl78_addr_space_pointer_mode): Likewise.
8417 (fl78_unwind_word_mode): Likewise.
8418 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
8420 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
8421 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
8422 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
8423 (mips_valid_pointer_mode): Likewise.
8424 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
8425 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
8426 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
8428 (ft32_addr_space_address_mode): Likewise.
8429 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
8430 scalar_int_mode instead of a machine_mode.
8431 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
8433 (m32c_addr_space_address_mode): Likewise.
8434 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
8435 (rs6000_eh_return_filter_mode): Likewise.
8436 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
8437 (rs6000_eh_return_filter_mode): Likewise.
8438 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
8439 (s390_libgcc_shift_count_mode): Likewise.
8440 (s390_unwind_word_mode): Likewise.
8441 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
8443 * target.def (mode_rep_extended): Likewise.
8444 (valid_pointer_mode): Likewise.
8445 (addr_space.valid_pointer_mode): Likewise.
8446 (eh_return_filter_mode): Return a scalar_int_mode rather than
8448 (libgcc_cmp_return_mode): Likewise.
8449 (libgcc_shift_count_mode): Likewise.
8450 (unwind_word_mode): Likewise.
8451 (addr_space.pointer_mode): Likewise.
8452 (addr_space.address_mode): Likewise.
8453 * doc/tm.texi: Regenerate.
8454 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
8456 (do_jump): Use scalar_int_mode for local variables.
8457 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
8458 rather than a machine_mode.
8459 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
8460 (scompare_loc_descriptor_wide): Likewise.
8461 (scompare_loc_descriptor_narrow): Likewise.
8462 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
8464 * except.c (sjlj_emit_dispatch_table): Likewise.
8465 (expand_builtin_eh_copy_values): Likewise.
8466 * explow.c (convert_memory_address_addr_space_1): Likewise.
8467 Take a scalar_int_mode rather than a machine_mode.
8468 (convert_memory_address_addr_space): Take a scalar_int_mode rather
8469 than a machine_mode.
8470 (memory_address_addr_space): Use scalar_int_mode for local variables.
8471 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
8472 rather than a machine_mode.
8473 * expmed.c (mask_rtx): Likewise.
8474 (init_expmed_one_conv): Likewise.
8475 (expand_mult_highpart_adjust): Likewise.
8476 (extract_high_half): Likewise.
8477 (expmed_mult_highpart_optab): Likewise.
8478 (expmed_mult_highpart): Likewise.
8479 (expand_smod_pow2): Likewise.
8480 (expand_sdiv_pow2): Likewise.
8481 (emit_store_flag_int): Likewise.
8482 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
8484 (extract_low_bits): Likewise.
8485 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
8487 * expr.c (pieces_addr::adjust): Likewise.
8488 (can_store_by_pieces): Likewise.
8489 (store_by_pieces): Likewise.
8490 (clear_by_pieces_1): Likewise.
8491 (expand_expr_addr_expr_1): Likewise.
8492 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
8493 (expand_expr_real_1): Likewise.
8494 (try_casesi): Likewise.
8495 * final.c (shorten_branches): Likewise.
8496 * fold-const.c (fold_convert_const_int_from_fixed): Change the
8497 type of "mode" to machine_mode.
8498 * internal-fn.c (expand_arith_overflow_result_store): Take a
8499 scalar_int_mode rather than a machine_mode.
8500 (expand_mul_overflow): Use scalar_int_mode for local variables.
8501 * loop-doloop.c (doloop_modify): Likewise.
8502 (doloop_optimize): Likewise.
8503 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
8504 than a machine_mode.
8505 (expand_doubleword_shift_condmove): Likewise.
8506 (expand_doubleword_shift): Likewise.
8507 (expand_doubleword_clz): Likewise.
8508 (expand_doubleword_popcount): Likewise.
8509 (expand_doubleword_parity): Likewise.
8510 (expand_absneg_bit): Use scalar_int_mode for local variables.
8511 (prepare_float_lib_cmp): Likewise.
8512 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
8513 rather than a machine_mode.
8514 (convert_memory_address_addr_space): Likewise.
8515 (get_mode_bounds): Likewise.
8516 (get_address_mode): Return a scalar_int_mode rather than a
8518 * rtlanal.c (get_address_mode): Likewise.
8519 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
8520 than a machine_mode.
8521 * targhooks.c (default_mode_rep_extended): Likewise.
8522 (default_valid_pointer_mode): Likewise.
8523 (default_addr_space_valid_pointer_mode): Likewise.
8524 (default_eh_return_filter_mode): Return a scalar_int_mode rather
8525 than a machine_mode.
8526 (default_libgcc_cmp_return_mode): Likewise.
8527 (default_libgcc_shift_count_mode): Likewise.
8528 (default_unwind_word_mode): Likewise.
8529 (default_addr_space_pointer_mode): Likewise.
8530 (default_addr_space_address_mode): Likewise.
8531 * targhooks.h (default_eh_return_filter_mode): Likewise.
8532 (default_libgcc_cmp_return_mode): Likewise.
8533 (default_libgcc_shift_count_mode): Likewise.
8534 (default_unwind_word_mode): Likewise.
8535 (default_addr_space_pointer_mode): Likewise.
8536 (default_addr_space_address_mode): Likewise.
8537 (default_mode_rep_extended): Take a scalar_int_mode rather than
8539 (default_valid_pointer_mode): Likewise.
8540 (default_addr_space_valid_pointer_mode): Likewise.
8541 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
8543 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
8544 rather than a machine_mode.
8545 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
8546 for local variables.
8547 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8548 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
8549 than a machine_mode.
8551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8552 Alan Hayward <alan.hayward@arm.com>
8553 David Sherwood <david.sherwood@arm.com>
8555 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
8556 the mode argument to scalar_int_mode.
8557 (do_jump_by_parts_zero_rtx): Likewise.
8558 (do_jump_by_parts_equality_rtx): Likewise.
8559 (do_jump_by_parts_greater): Take a mode argument.
8560 (do_jump_by_parts_equality): Likewise.
8561 (do_jump_1): Update calls accordingly.
8563 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8564 Alan Hayward <alan.hayward@arm.com>
8565 David Sherwood <david.sherwood@arm.com>
8567 * is-a.h (safe_dyn_cast): New function.
8568 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
8569 (jump_table_for_label): Likewise.
8570 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
8571 instead of an rtx_insn *.
8572 (shorten_branches): Use dyn_cast instead of LABEL_P and
8573 JUMP_TABLE_DATA_P. Use jump_table_for_label and
8574 rtx_jump_table_data::get_data_mode.
8575 (final_scan_insn): Likewise.
8577 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8578 Alan Hayward <alan.hayward@arm.com>
8579 David Sherwood <david.sherwood@arm.com>
8581 * combine.c (try_combine): Use is_a <scalar_int_mode> when
8582 trying to combine a full-register integer set with a subreg
8585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8586 Alan Hayward <alan.hayward@arm.com>
8587 David Sherwood <david.sherwood@arm.com>
8589 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
8590 that is always either address_mode or pointer_mode.
8592 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8593 Alan Hayward <alan.hayward@arm.com>
8594 David Sherwood <david.sherwood@arm.com>
8596 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
8597 when the two are known to be equal.
8599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8600 Alan Hayward <alan.hayward@arm.com>
8601 David Sherwood <david.sherwood@arm.com>
8603 * simplify-rtx.c (simplify_const_unary_operation): Use
8604 is_a <scalar_int_mode> instead of checking for a nonzero
8605 precision. Forcibly convert op_mode to a scalar_int_mode
8606 in that case. More clearly differentiate the operand and
8607 result modes and use the former when deciding what the value
8608 of a count-bits operation should be. Use is_int_mode instead
8609 of checking for a MODE_INT. Remove redundant check for whether
8610 this mode has a zero precision.
8612 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8613 Alan Hayward <alan.hayward@arm.com>
8614 David Sherwood <david.sherwood@arm.com>
8616 * optabs.c (widen_leading): Change the type of the mode argument
8617 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
8618 (widen_bswap): Likewise.
8619 (expand_parity): Likewise.
8620 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
8621 (expand_ffs): Likewise.
8622 (epand_unop): Check for scalar integer modes before calling the
8625 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8626 Alan Hayward <alan.hayward@arm.com>
8627 David Sherwood <david.sherwood@arm.com>
8629 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
8631 (expand_expr_real_1): Update call accordingly.
8633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8634 Alan Hayward <alan.hayward@arm.com>
8635 David Sherwood <david.sherwood@arm.com>
8637 * expmed.c (store_bit_field_using_insv): Add op0_mode and
8638 value_mode arguments. Use scalar_int_mode internally.
8639 (store_bit_field_1): Rename the new integer mode from imode
8640 to op0_mode and use it instead of GET_MODE (op0). Update calls
8641 to store_split_bit_field, store_bit_field_using_insv and
8642 store_fixed_bit_field.
8643 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
8644 Use scalar_int_mode internally. Use a bit count rather than a mode
8645 when calculating the largest bit size for get_best_mode.
8646 Update calls to store_split_bit_field and store_fixed_bit_field_1.
8647 (store_fixed_bit_field_1): Add mode and value_mode arguments.
8648 Remove assertion that OP0 has a scalar integer mode.
8649 (store_split_bit_field): Add op0_mode and value_mode arguments.
8650 Update calls to extract_fixed_bit_field.
8651 (extract_bit_field_using_extv): Add an op0_mode argument.
8652 Use scalar_int_mode internally.
8653 (extract_bit_field_1): Rename the new integer mode from imode to
8654 op0_mode and use it instead of GET_MODE (op0). Update calls to
8655 extract_split_bit_field, extract_bit_field_using_extv and
8656 extract_fixed_bit_field.
8657 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
8658 to extract_split_bit_field and extract_fixed_bit_field_1.
8659 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
8660 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
8662 (extract_split_bit_field): Add an op0_mode argument. Update call
8663 to extract_fixed_bit_field.
8665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8666 Alan Hayward <alan.hayward@arm.com>
8667 David Sherwood <david.sherwood@arm.com>
8669 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
8670 * explow.c (hard_function_value): Likewise.
8671 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
8672 convert_to_mode call outside the loop.
8673 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
8674 for the mode iterator. Require the mode specified by max_pieces
8676 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
8678 (copy_blkmode_to_reg): Likewise.
8679 (set_storage_via_setmem): Likewise.
8680 * optabs.c (prepare_cmp_insn): Likewise.
8681 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8682 * stor-layout.c (finish_bitfield_representative): Likewise.
8684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8685 Alan Hayward <alan.hayward@arm.com>
8686 David Sherwood <david.sherwood@arm.com>
8688 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
8689 * expr.c (convert_move): Use them.
8690 (convert_modes): Likewise.
8691 (store_expr_with_bounds): Likewise.
8693 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8694 Alan Hayward <alan.hayward@arm.com>
8695 David Sherwood <david.sherwood@arm.com>
8697 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
8698 parameter for the mode of "x". Remove the "known_x", "known_mode"
8699 and "known_ret" arguments. Change the type of the mode argument
8701 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
8702 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
8703 (reg_num_sign_bit_copies_for_combine): Likewise.
8704 * rtlanal.c (nonzero_bits1): Likewise.
8705 (num_sign_bit_copies1): Likewise.
8706 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
8707 (reg_num_sign_bit_copies_general): Likewise.
8708 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
8709 (reg_nonzero_bits_general): Likewise.
8711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8712 Alan Hayward <alan.hayward@arm.com>
8713 David Sherwood <david.sherwood@arm.com>
8715 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
8716 than in subroutines. Return 1 for non-integer modes.
8717 (cached_num_sign_bit_copies): Change the type of the mode parameter
8719 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
8720 classes. Handle CONST_INT_P first and then check whether X also
8721 has a scalar integer mode. Check the same thing for inner registers
8722 of a SUBREG and for values that are being extended or truncated.
8724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8725 Alan Hayward <alan.hayward@arm.com>
8726 David Sherwood <david.sherwood@arm.com>
8728 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
8729 in subroutines. Return the mode mask for non-integer modes.
8730 (cached_nonzero_bits): Change the type of the mode parameter
8732 (nonzero_bits1): Likewise. Remove early exit for other mode
8733 classes. Handle CONST_INT_P first and then check whether X
8734 also has a scalar integer mode.
8736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8737 Alan Hayward <alan.hayward@arm.com>
8738 David Sherwood <david.sherwood@arm.com>
8740 * expr.c (widest_int_mode_for_size): Make the comment match the code.
8741 Return a scalar_int_mode and assert that the size is greater than
8743 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
8744 (op_by_pieces_d::op_by_pieces_d): Likewise.
8745 (op_by_pieces_d::run): Likewise.
8746 (can_store_by_pieces): Likewise.
8748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8749 Alan Hayward <alan.hayward@arm.com>
8750 David Sherwood <david.sherwood@arm.com>
8752 * combine.c (extract_left_shift): Add a mode argument and update
8754 (make_compound_operation_int): Change the type of the mode parameter
8755 to scalar_int_mode and update the call to extract_left_shift.
8757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8758 Alan Hayward <alan.hayward@arm.com>
8759 David Sherwood <david.sherwood@arm.com>
8761 * combine.c (simplify_and_const_int): Change the type of the mode
8762 parameter to scalar_int_mode.
8763 (simplify_and_const_int_1): Likewise. Update recursive call.
8765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8766 Alan Hayward <alan.hayward@arm.com>
8767 David Sherwood <david.sherwood@arm.com>
8769 * combine.c (simplify_compare_const): Check that the mode is a
8770 scalar_int_mode (rather than VOIDmode) before testing its
8772 (simplify_comparison): Move COMPARISON_P handling out of the
8773 loop and restrict the latter part of the loop to scalar_int_modes.
8774 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
8775 and when considering SUBREG_REGs. Use is_int_mode instead of
8776 checking GET_MODE_CLASS against MODE_INT.
8778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8779 Alan Hayward <alan.hayward@arm.com>
8780 David Sherwood <david.sherwood@arm.com>
8782 * combine.c (try_widen_shift_mode): Move check for equal modes to...
8783 (simplify_shift_const_1): ...here. Use scalar_int_mode for
8784 shift_unit_mode and for modes involved in scalar shifts.
8786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8787 Alan Hayward <alan.hayward@arm.com>
8788 David Sherwood <david.sherwood@arm.com>
8790 * combine.c (force_int_to_mode): New function, split out from...
8791 (force_to_mode): ...here. Keep xmode up-to-date and use it
8792 instead of GET_MODE (x).
8794 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8795 Alan Hayward <alan.hayward@arm.com>
8796 David Sherwood <david.sherwood@arm.com>
8798 * optabs-query.h (extraction_insn::struct_mode): Change type to
8799 opt_scalar_int_mode and update comment.
8800 (extraction_insn::field_mode): Change type to scalar_int_mode.
8801 (extraction_insn::pos_mode): Likewise.
8802 * combine.c (make_extraction): Update accordingly.
8803 * optabs-query.c (get_traditional_extraction_insn): Likewise.
8804 (get_optab_extraction_insn): Likewise.
8805 * recog.c (simplify_while_replacing): Likewise.
8806 * expmed.c (narrow_bit_field_mem): Change the type of the mode
8807 parameter to opt_scalar_int_mode.
8809 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8810 Alan Hayward <alan.hayward@arm.com>
8811 David Sherwood <david.sherwood@arm.com>
8813 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
8814 to a scalar_int_mode instead of a machine_mode.
8815 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
8816 (get_best_mode): Return a boolean and use a pointer argument to store
8817 the selected mode. Replace the limit mode parameter with a bit limit.
8818 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
8819 for the values returned by bit_field_mode_iterator::next_mode.
8820 (store_bit_field): Update call to get_best_mode.
8821 (store_fixed_bit_field): Likewise.
8822 (extract_fixed_bit_field): Likewise.
8823 * expr.c (optimize_bitfield_assignment_op): Likewise.
8824 * fold-const.c (optimize_bit_field_compare): Likewise.
8825 (fold_truth_andor_1): Likewise.
8826 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
8827 Update for new type of m_mode.
8828 (get_best_mode): As above.
8830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8831 Alan Hayward <alan.hayward@arm.com>
8832 David Sherwood <david.sherwood@arm.com>
8834 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
8835 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
8836 (store_bit_field): Check is_a <scalar_int_mode> before calling
8837 strict_volatile_bitfield_p.
8838 (extract_bit_field): Likewise.
8840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8841 Alan Hayward <alan.hayward@arm.com>
8842 David Sherwood <david.sherwood@arm.com>
8844 * target.def (cstore_mode): Return a scalar_int_mode.
8845 * doc/tm.texi: Regenerate.
8846 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
8847 * targhooks.h (default_cstore_mode): Likewise.
8848 * targhooks.c (default_cstore_mode): Likewise, using a forced
8850 * expmed.c (emit_cstore): Expect the target of the cstore to be
8853 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8854 Alan Hayward <alan.hayward@arm.com>
8855 David Sherwood <david.sherwood@arm.com>
8857 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
8859 (niter_desc): Likewise mode.
8860 (iv_analyze): Add a mode parameter.
8862 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
8863 and change its type to scalar_int_mode.
8864 * loop-iv.c: Update commentary at head of file.
8865 (iv_constant): Pass the mode paraeter before the rtx it describes
8866 and change its type to scalar_int_mode. Remove VOIDmode handling.
8867 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
8868 (iv_extend): Likewise.
8869 (shorten_into_mode): Likewise.
8870 (iv_add): Use scalar_int_mode.
8871 (iv_mult): Likewise.
8872 (iv_shift): Likewise.
8873 (canonicalize_iv_subregs): Likewise.
8874 (get_biv_step_1): Pass the outer_mode parameter before the rtx
8875 it describes and change its mode to scalar_int_mode. Also change
8876 the type of the returned inner_mode to scalar_int_mode.
8877 (get_biv_step): Likewise, turning outer_mode from a pointer
8878 into a direct parameter. Update call to get_biv_step_1.
8879 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
8880 iv_constant and get_biv_step.
8881 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
8882 and change its type to scalar_int_mode. Don't initialise iv->mode
8883 to VOIDmode and remove later checks for its still being VOIDmode.
8884 Update calls to iv_analyze_op and iv_analyze_expr. Check
8885 is_a <scalar_int_mode> when changing the mode under consideration.
8886 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
8887 Update call to iv_analyze_expr.
8888 (iv_analyze_op): Add a mode parameter. Reject subregs whose
8889 inner register is not also a scalar_int_mode. Update call to
8891 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
8892 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
8893 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
8894 separate mode class checks. Update calls to iv_analyze. Remove
8895 fix-up of VOIDmodes after iv_analyze_biv.
8896 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
8897 don't have a scalar_int_mode. Update call to biv_p.
8899 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8900 Alan Hayward <alan.hayward@arm.com>
8901 David Sherwood <david.sherwood@arm.com>
8903 * cfgexpand.c (convert_debug_memory_address): Use
8904 as_a <scalar_int_mode>.
8905 * combine.c (expand_compound_operation): Likewise.
8906 (make_extraction): Likewise.
8907 (change_zero_ext): Likewise.
8908 (simplify_comparison): Likewise.
8909 * cse.c (cse_insn): Likewise.
8910 * dwarf2out.c (minmax_loc_descriptor): Likewise.
8911 (mem_loc_descriptor): Likewise.
8912 (loc_descriptor): Likewise.
8913 * expmed.c (init_expmed_one_mode): Likewise.
8914 (synth_mult): Likewise.
8915 (emit_store_flag_1): Likewise.
8916 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
8917 of a comparison with size.
8918 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
8919 (reduce_to_bit_field_precision): Likewise.
8920 * function.c (expand_function_end): Likewise.
8921 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
8922 * loop-doloop.c (doloop_modify): Likewise.
8923 * optabs.c (expand_binop): Likewise.
8924 (expand_unop): Likewise.
8925 (expand_copysign_absneg): Likewise.
8926 (prepare_cmp_insn): Likewise.
8927 (maybe_legitimize_operand): Likewise.
8928 * recog.c (const_scalar_int_operand): Likewise.
8929 * rtlanal.c (get_address_mode): Likewise.
8930 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8931 (simplify_cond_clz_ctz): Likewise.
8932 * tree-nested.c (get_nl_goto_field): Likewise.
8933 * tree.c (build_vector_type_for_mode): Likewise.
8934 * var-tracking.c (use_narrower_mode): Likewise.
8936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8937 Alan Hayward <alan.hayward@arm.com>
8938 David Sherwood <david.sherwood@arm.com>
8940 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
8941 * builtins.c (expand_builtin_signbit): Use it.
8942 * cfgexpand.c (expand_debug_expr): Likewise.
8943 * dojump.c (do_jump): Likewise.
8944 (do_compare_and_jump): Likewise.
8945 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
8946 * expmed.c (make_tree): Likewise.
8947 * expr.c (expand_expr_real_2): Likewise.
8948 (expand_expr_real_1): Likewise.
8949 (try_casesi): Likewise.
8950 * fold-const-call.c (fold_const_call_ss): Likewise.
8951 * fold-const.c (unextend): Likewise.
8952 (extract_muldiv_1): Likewise.
8953 (fold_single_bit_test): Likewise.
8954 (native_encode_int): Likewise.
8955 (native_encode_string): Likewise.
8956 (native_interpret_int): Likewise.
8957 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
8958 * internal-fn.c (expand_addsub_overflow): Likewise.
8959 (expand_neg_overflow): Likewise.
8960 (expand_mul_overflow): Likewise.
8961 (expand_arith_overflow): Likewise.
8962 * match.pd: Likewise.
8963 * stor-layout.c (layout_type): Likewise.
8964 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
8965 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
8966 * tree-ssanames.c (get_range_info): Likewise.
8967 * tree-switch-conversion.c (array_value_type) Likewise.
8968 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
8969 (vect_recog_divmod_pattern): Likewise.
8970 (vect_recog_mixed_size_cond_pattern): Likewise.
8971 * tree-vrp.c (extract_range_basic): Likewise.
8972 (simplify_float_conversion_using_ranges): Likewise.
8973 * tree.c (int_fits_type_p): Likewise.
8974 * ubsan.c (instrument_bool_enum_load): Likewise.
8975 * varasm.c (mergeable_string_section): Likewise.
8976 (narrowing_initializer_constant_valid_p): Likewise.
8977 (output_constant): Likewise.
8979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8980 Alan Hayward <alan.hayward@arm.com>
8981 David Sherwood <david.sherwood@arm.com>
8983 * machmode.h (NARROWEST_INT_MODE): New macro.
8984 * expr.c (alignment_for_piecewise_move): Use it instead of
8985 GET_CLASS_NARROWEST_MODE (MODE_INT).
8986 (push_block): Likewise.
8987 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
8989 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8992 Alan Hayward <alan.hayward@arm.com>
8993 David Sherwood <david.sherwood@arm.com>
8995 * postreload.c (move2add_valid_value_p): Change the type of the
8996 mode parameter to scalar_int_mode.
8997 (move2add_use_add2_insn): Add a mode parameter and use it instead
8999 (move2add_use_add3_insn): Likewise.
9000 (reload_cse_move2add): Update accordingly.
9002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9003 Alan Hayward <alan.hayward@arm.com>
9004 David Sherwood <david.sherwood@arm.com>
9006 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
9008 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
9009 * optabs.c (expand_unop): Likewise.
9011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9012 Alan Hayward <alan.hayward@arm.com>
9013 David Sherwood <david.sherwood@arm.com>
9015 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
9016 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
9017 (popcount_loc_descriptor): Likewise.
9018 (bswap_loc_descriptor): Likewise.
9019 (rotate_loc_descriptor): Likewise.
9020 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
9021 calling the functions above.
9023 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9024 Alan Hayward <alan.hayward@arm.com>
9025 David Sherwood <david.sherwood@arm.com>
9027 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
9029 (try_combine): Likewise.
9030 (simplify_if_then_else): Likewise.
9031 * cse.c (cse_insn): Likewise.
9032 * dwarf2out.c (mem_loc_descriptor): Likewise.
9033 * emit-rtl.c (gen_lowpart_common): Likewise.
9034 * simplify-rtx.c (simplify_truncation): Likewise.
9035 (simplify_binary_operation_1): Likewise.
9036 (simplify_const_relational_operation): Likewise.
9037 (simplify_ternary_operation): Likewise.
9038 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
9040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9041 Alan Hayward <alan.hayward@arm.com>
9042 David Sherwood <david.sherwood@arm.com>
9044 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
9045 * reload.c (push_reload): Likewise.
9046 (find_reloads): Likewise.
9048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9049 Alan Hayward <alan.hayward@arm.com>
9050 David Sherwood <david.sherwood@arm.com>
9052 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
9053 (make_compound_operation_int): Likewise.
9054 (change_zero_ext): Likewise.
9055 * expr.c (convert_move): Likewise.
9056 (convert_modes): Likewise.
9057 * fwprop.c (forward_propagate_subreg): Likewise.
9058 * loop-iv.c (get_biv_step_1): Likewise.
9059 * optabs.c (widen_operand): Likewise.
9060 * postreload.c (move2add_valid_value_p): Likewise.
9061 * recog.c (simplify_while_replacing): Likewise.
9062 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9063 (simplify_binary_operation_1): Likewise. Remove redundant
9064 mode equality check.
9066 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9067 Alan Hayward <alan.hayward@arm.com>
9068 David Sherwood <david.sherwood@arm.com>
9070 * combine.c (combine_simplify_rtx): Add checks for
9071 is_a <scalar_int_mode>.
9072 (simplify_if_then_else): Likewise.
9073 (make_field_assignment): Likewise.
9074 (simplify_comparison): Likewise.
9075 * ifcvt.c (noce_try_bitop): Likewise.
9076 * loop-invariant.c (canonicalize_address_mult): Likewise.
9077 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9079 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9080 Alan Hayward <alan.hayward@arm.com>
9081 David Sherwood <david.sherwood@arm.com>
9083 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
9084 is_a <scalar_int_mode> instead of != BLKmode.
9086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9087 Alan Hayward <alan.hayward@arm.com>
9088 David Sherwood <david.sherwood@arm.com>
9090 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
9091 instead of != VOIDmode.
9092 * combine.c (if_then_else_cond): Likewise.
9093 (change_zero_ext): Likewise.
9094 * dwarf2out.c (mem_loc_descriptor): Likewise.
9095 (loc_descriptor): Likewise.
9096 * rtlanal.c (canonicalize_condition): Likewise.
9097 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
9099 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9100 Alan Hayward <alan.hayward@arm.com>
9101 David Sherwood <david.sherwood@arm.com>
9103 * simplify-rtx.c (simplify_binary_operation_1): Use
9104 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
9106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9107 Alan Hayward <alan.hayward@arm.com>
9108 David Sherwood <david.sherwood@arm.com>
9110 * wide-int.h (int_traits<unsigned char>) New class.
9111 (int_traits<unsigned short>) Likewise.
9112 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
9113 Use GET_MODE_UNIT_PRECISION and remove redundant test for
9115 * combine.c (set_nonzero_bits_and_sign_copies): Use
9116 is_a <scalar_int_mode>.
9117 (find_split_point): Likewise.
9118 (combine_simplify_rtx): Likewise.
9119 (simplify_logical): Likewise.
9120 (expand_compound_operation): Likewise.
9121 (expand_field_assignment): Likewise.
9122 (make_compound_operation): Likewise.
9123 (extended_count): Likewise.
9124 (change_zero_ext): Likewise.
9125 (simplify_comparison): Likewise.
9126 * dwarf2out.c (scompare_loc_descriptor): Likewise.
9127 (ucompare_loc_descriptor): Likewise.
9128 (minmax_loc_descriptor): Likewise.
9129 (mem_loc_descriptor): Likewise.
9130 (loc_descriptor): Likewise.
9131 * expmed.c (init_expmed_one_mode): Likewise.
9132 * lra-constraints.c (lra_constraint_offset): Likewise.
9133 * optabs.c (prepare_libcall_arg): Likewise.
9134 * postreload.c (move2add_note_store): Likewise.
9135 * reload.c (operands_match_p): Likewise.
9136 * rtl.h (load_extend_op): Likewise.
9137 * rtlhooks.c (gen_lowpart_general): Likewise.
9138 * simplify-rtx.c (simplify_truncation): Likewise.
9139 (simplify_unary_operation_1): Likewise.
9140 (simplify_binary_operation_1): Likewise.
9141 (simplify_const_binary_operation): Likewise.
9142 (simplify_const_relational_operation): Likewise.
9143 (simplify_subreg): Likewise.
9144 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9145 * var-tracking.c (adjust_mems): Likewise.
9146 (prepare_call_arguments): Likewise.
9148 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9149 Alan Hayward <alan.hayward@arm.com>
9150 David Sherwood <david.sherwood@arm.com>
9152 * machmode.h (is_int_mode): New fuction.
9153 * combine.c (find_split_point): Use it.
9154 (combine_simplify_rtx): Likewise.
9155 (simplify_if_then_else): Likewise.
9156 (simplify_set): Likewise.
9157 (simplify_shift_const_1): Likewise.
9158 (simplify_comparison): Likewise.
9159 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
9160 * cse.c (notreg_cost): Likewise.
9161 (cse_insn): Likewise.
9162 * cselib.c (cselib_lookup_1): Likewise.
9163 * dojump.c (do_jump_1): Likewise.
9164 (do_compare_rtx_and_jump): Likewise.
9165 * dse.c (get_call_args): Likewise.
9166 * dwarf2out.c (rtl_for_decl_init): Likewise.
9167 (native_encode_initializer): Likewise.
9168 * expmed.c (emit_store_flag_1): Likewise.
9169 (emit_store_flag): Likewise.
9170 * expr.c (convert_modes): Likewise.
9171 (store_field): Likewise.
9172 (expand_expr_real_1): Likewise.
9173 * fold-const.c (fold_read_from_constant_string): Likewise.
9174 * gimple-ssa-sprintf.c (get_format_string): Likewise.
9175 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
9176 * optabs.c (expand_binop): Likewise.
9177 (expand_unop): Likewise.
9178 (expand_abs_nojump): Likewise.
9179 (expand_one_cmpl_abs_nojump): Likewise.
9180 * simplify-rtx.c (mode_signbit_p): Likewise.
9181 (val_signbit_p): Likewise.
9182 (val_signbit_known_set_p): Likewise.
9183 (val_signbit_known_clear_p): Likewise.
9184 (simplify_relational_operation_1): Likewise.
9185 * tree.c (vector_type_mode): Likewise.
9187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9188 Alan Hayward <alan.hayward@arm.com>
9189 David Sherwood <david.sherwood@arm.com>
9191 * machmode.h (smallest_mode_for_size): Fix formatting.
9192 (smallest_int_mode_for_size): New function.
9193 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
9194 instead of smallest_mode_for_size.
9195 * combine.c (make_extraction): Likewise.
9196 * config/arc/arc.c (arc_expand_movmem): Likewise.
9197 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
9198 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9199 * config/s390/s390.c (s390_expand_insv): Likewise.
9200 * config/sparc/sparc.c (assign_int_registers): Likewise.
9201 * config/spu/spu.c (spu_function_value): Likewise.
9202 (spu_function_arg): Likewise.
9203 * coverage.c (get_gcov_type): Likewise.
9204 (get_gcov_unsigned_t): Likewise.
9205 * dse.c (find_shift_sequence): Likewise.
9206 * expmed.c (store_bit_field_1): Likewise.
9207 * expr.c (convert_move): Likewise.
9208 (store_field): Likewise.
9209 * internal-fn.c (expand_arith_overflow): Likewise.
9210 * optabs-query.c (get_best_extraction_insn): Likewise.
9211 * optabs.c (expand_twoval_binop_libfunc): Likewise.
9212 * stor-layout.c (layout_type): Likewise.
9213 (initialize_sizetypes): Likewise.
9214 * targhooks.c (default_get_mask_mode): Likewise.
9215 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
9217 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9218 Alan Hayward <alan.hayward@arm.com>
9219 David Sherwood <david.sherwood@arm.com>
9221 * machmode.h (opt_mode::else_blk): New function.
9222 (int_mode_for_mode): Declare.
9223 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
9224 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
9226 * cfgexpand.c (expand_debug_expr): Likewise.
9227 * combine.c (gen_lowpart_or_truncate): Likewise.
9228 (gen_lowpart_for_combine): Likewise.
9229 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
9230 * config/avr/avr.c (avr_to_int_mode): Likewise.
9231 (avr_out_plus_1): Likewise.
9232 (avr_out_plus): Likewise.
9233 (avr_out_round): Likewise.
9234 * config/i386/i386.c (ix86_split_to_parts): Likewise.
9235 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9236 (s390_expand_vcond): Likewise.
9237 * config/spu/spu.c (spu_split_immediate): Likewise.
9238 (spu_expand_mov): Likewise.
9239 * dse.c (get_stored_val): Likewise.
9240 * expmed.c (store_bit_field_1): Likewise.
9241 (convert_extracted_bit_field): Use int_mode_for_mode instead of
9243 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
9244 (extract_low_bits): Likewise.
9245 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
9246 handling rather than repeating the check.
9247 (emit_group_store): Likewise.
9248 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
9249 * optabs.c (expand_absneg_bit): Likewise.
9250 (expand_copysign_absneg): Likewise.
9251 (expand_copysign_bit): Likewise.
9252 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
9253 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
9254 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
9255 * var-tracking.c (prepare_call_arguments): Likewise.
9256 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
9257 int_mode_for_mode instead of mode_for_size.
9258 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9261 Alan Hayward <alan.hayward@arm.com>
9262 David Sherwood <david.sherwood@arm.com>
9264 * machmode.h (int_mode_for_size): New function.
9265 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
9266 instead of mode_for_size.
9267 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
9269 * combine.c (expand_field_assignment): Use int_mode_for_size
9270 instead of mode_for_size.
9271 (make_extraction): Likewise.
9272 (simplify_shift_const_1): Likewise.
9273 (simplify_comparison): Likewise.
9274 * dojump.c (do_jump): Likewise.
9275 * dwarf2out.c (mem_loc_descriptor): Likewise.
9276 * emit-rtl.c (init_derived_machine_modes): Likewise.
9277 * expmed.c (flip_storage_order): Likewise.
9278 (convert_extracted_bit_field): Likewise.
9279 * expr.c (copy_blkmode_from_reg): Likewise.
9280 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
9281 * internal-fn.c (expand_mul_overflow): Likewise.
9282 * lower-subreg.c (simple_move): Likewise.
9283 * optabs-libfuncs.c (init_optabs): Likewise.
9284 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9285 * tree.c (vector_type_mode): Likewise.
9286 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
9287 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
9288 * tree-vect-generic.c (expand_vector_parallel): Likewise.
9289 * tree-vect-stmts.c (vectorizable_load): Likewise.
9290 (vectorizable_store): Likewise.
9292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9293 Alan Hayward <alan.hayward@arm.com>
9294 David Sherwood <david.sherwood@arm.com>
9296 * coretypes.h (pod_mode): New type.
9297 (scalar_int_mode_pod): New typedef.
9298 * machmode.h (pod_mode): New class.
9299 (int_n_data_t::m): Change type to scalar_int_mode_pod.
9300 * genmodes.c (emit_mode_int_n): Update accordingly.
9301 * lower-subreg.h (target_lower_subreg): Change type to
9302 scalar_int_mode_pod.
9303 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
9304 scalar_int_mode_pod.
9306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9307 Alan Hayward <alan.hayward@arm.com>
9308 David Sherwood <david.sherwood@arm.com>
9310 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
9311 machine_mode to scalar_int_mode.
9312 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
9313 (rs6000_option_override_internal): Remove cast to int.
9314 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
9315 machine_mode to scalar_int_mode.
9316 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
9317 (rs6000_option_override_internal): Remove cast to int.
9318 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
9319 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
9321 * config/s390/s390.c (s390_expand_builtin): Likewise.
9322 * coretypes.h (scalar_int_mode): New type.
9323 (opt_scalar_int_mode): New typedef.
9324 * machmode.h (scalar_int_mode): New class.
9325 (scalar_int_mode::includes_p): New function.
9326 (byte_mode): Change type to scalar_int_mode.
9327 (word_mode): Likewise.
9328 (ptr_mode): Likewise.
9329 * emit-rtl.c (byte_mode): Likewise.
9330 (word_mode): Likewise.
9331 (ptr_mode): Likewise.
9332 (init_derived_machine_modes): Update accordingly.
9333 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
9334 and MODE_PARTIAL_INT.
9335 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
9336 opt_scalar_int_mode.
9338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9339 Alan Hayward <alan.hayward@arm.com>
9340 David Sherwood <david.sherwood@arm.com>
9342 * target.def (libgcc_floating_mode_supported_p): Take a
9344 * doc/tm.texi: Regenerate.
9345 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
9347 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
9348 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
9351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9352 Alan Hayward <alan.hayward@arm.com>
9353 David Sherwood <david.sherwood@arm.com>
9355 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
9356 * doc/tm.texi: Regenerate.
9357 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
9358 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
9359 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
9360 * targhooks.h (default_floatn_mode): Likewise.
9361 * targhooks.c (default_floatn_mode): Likewise.
9362 * tree.c (build_common_tree_nodes): Update accordingly.
9364 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9365 Alan Hayward <alan.hayward@arm.com>
9366 David Sherwood <david.sherwood@arm.com>
9368 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
9369 (mode_iterator::iterate_p): Likewise.
9370 (mode_iterator::get_wider): Likewise.
9371 * expr.c (init_expr_target): Use opt_scalar_float_mode.
9373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9374 Alan Hayward <alan.hayward@arm.com>
9375 David Sherwood <david.sherwood@arm.com>
9377 * coretypes.h (opt_scalar_float_mode): New typedef.
9378 * machmode.h (float_mode_for_size): New function.
9379 * emit-rtl.c (double_mode): Delete.
9380 (init_emit_once): Use float_mode_for_size.
9381 * stor-layout.c (layout_type): Likewise.
9382 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
9384 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9385 Alan Hayward <alan.hayward@arm.com>
9386 David Sherwood <david.sherwood@arm.com>
9388 * output.h (assemble_real): Take a scalar_float_mode.
9389 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
9390 * config/arm/arm.md (consttable_4): Likewise.
9391 (consttable_8): Likewise.
9392 (consttable_16): Likewise.
9393 * config/mips/mips.md (consttable_float): Likewise.
9394 * config/s390/s390.c (s390_output_pool_entry): Likewise.
9395 * varasm.c (assemble_real): Take a scalar_float_mode.
9396 (output_constant_pool_2): Update accordingly.
9397 (output_constant): Likewise.
9399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9400 Alan Hayward <alan.hayward@arm.com>
9401 David Sherwood <david.sherwood@arm.com>
9403 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
9404 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
9405 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
9406 (native_encode_real): Likewise.
9407 (native_interpret_real): Likewise.
9408 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
9409 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9412 Alan Hayward <alan.hayward@arm.com>
9413 David Sherwood <david.sherwood@arm.com>
9415 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
9416 <scalar_float_mode>. Simplify.
9417 (gen_extend_conv_libfunc): Likewise.
9419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9420 Alan Hayward <alan.hayward@arm.com>
9421 David Sherwood <david.sherwood@arm.com>
9423 * coretypes.h (scalar_float_mode): New type.
9424 * machmode.h (mode_traits::from_int): Use machine_mode if
9425 USE_ENUM_MODES is defined.
9426 (is_a): New function.
9428 (dyn_cast): Likewise.
9429 (scalar_float_mode): New class.
9430 (scalar_float_mode::includes_p): New function.
9431 (is_float_mode): Likewise.
9432 * gdbhooks.py (MachineModePrinter): New class.
9433 (build_pretty_printer): Use it for scalar_float_mode.
9434 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
9435 (format_helper::format_helper): Turn into a template.
9436 * genmodes.c (get_mode_class): New function.
9437 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
9438 or machine_mode if none.
9439 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
9440 as_a <scalar_float_mode>.
9441 * dwarf2out.c (mem_loc_descriptor): Likewise.
9442 (insert_float): Likewise.
9443 (add_const_value_attribute): Likewise.
9444 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9445 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
9446 (expand_unop): Update accordingly.
9447 (expand_abs_nojump): Likewise.
9448 (expand_copysign_absneg): Take a scalar_float_mode.
9449 (expand_copysign_bit): Likewise.
9450 (expand_copysign): Update accordingly.
9452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9453 Alan Hayward <alan.hayward@arm.com>
9454 David Sherwood <david.sherwood@arm.com>
9456 * coretypes.h (opt_mode): New class.
9457 * machmode.h (opt_mode): Likewise.
9458 (opt_mode::else_void): New function.
9459 (opt_mode::require): Likewise.
9460 (opt_mode::exists): Likewise.
9461 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
9462 (GET_MODE_2XWIDER_MODE): Likewise.
9463 (mode_iterator::get_wider): Update accordingly.
9464 (mode_iterator::get_2xwider): Likewise.
9465 (mode_iterator::get_known_wider): Likewise, turning into a template.
9466 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
9467 forcing a wider mode to exist.
9468 * config/cr16/cr16.h (LONG_REG_P): Likewise.
9469 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9470 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
9471 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
9472 * lower-subreg.c (init_lower_subreg): Likewise.
9473 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
9474 on the final iteration.
9475 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
9476 a wider mode exists before asking for a move pattern.
9477 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
9478 forcing a wider mode to exist.
9479 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
9480 returning false if no such mode exists.
9481 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
9482 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
9483 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
9484 Avoid checking for a MODE_INT if we already know the mode is not a
9486 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
9487 forcing a wider mode to exist.
9488 (expmed_mult_highpart_optab): Likewise.
9489 (expmed_mult_highpart): Likewise.
9490 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
9492 * lto-streamer-in.c (lto_input_mode_table): Likewise.
9493 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
9494 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9495 * internal-fn.c (expand_mul_overflow): Update use of
9496 GET_MODE_2XWIDER_MODE.
9497 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9498 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
9499 GET_MODE_WIDER_MODE.
9500 (convert_plusminus_to_widen): Likewise.
9501 * tree-switch-conversion.c (array_value_type): Likewise.
9502 * var-tracking.c (emit_note_insn_var_location): Likewise.
9503 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9504 Return false inside rather than outside the loop if no wider mode
9506 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
9507 and GET_MODE_2XWIDER_MODE
9508 (can_compare_p): Use else_void.
9509 * gdbhooks.py (OptMachineModePrinter): New class.
9510 (build_pretty_printer): Use it for opt_mode.
9512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9513 Alan Hayward <alan.hayward@arm.com>
9514 David Sherwood <david.sherwood@arm.com>
9516 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
9517 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
9519 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9520 Alan Hayward <alan.hayward@arm.com>
9521 David Sherwood <david.sherwood@arm.com>
9523 * machmode.h (mode_traits): New structure.
9524 (get_narrowest_mode): New function.
9525 (mode_iterator::start): Likewise.
9526 (mode_iterator::iterate_p): Likewise.
9527 (mode_iterator::get_wider): Likewise.
9528 (mode_iterator::get_known_wider): Likewise.
9529 (mode_iterator::get_2xwider): Likewise.
9530 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
9531 (FOR_EACH_MODE): Likewise.
9532 (FOR_EACH_MODE_FROM): Likewise.
9533 (FOR_EACH_MODE_UNTIL): Likewise.
9534 (FOR_EACH_WIDER_MODE): Likewise.
9535 (FOR_EACH_2XWIDER_MODE): Likewise.
9536 * builtins.c (expand_builtin_strlen): Use new mode iterators.
9537 * combine.c (simplify_comparison): Likewise
9538 * config/i386/i386.c (type_natural_mode): Likewise.
9539 * cse.c (cse_insn): Likewise.
9540 * dse.c (find_shift_sequence): Likewise.
9541 * emit-rtl.c (init_derived_machine_modes): Likewise.
9542 (init_emit_once): Likewise.
9543 * explow.c (hard_function_value): Likewise.
9544 * expmed.c (extract_fixed_bit_field_1): Likewise.
9545 (extract_bit_field_1): Likewise.
9546 (expand_divmod): Likewise.
9547 (emit_store_flag_1): Likewise.
9548 * expr.c (init_expr_target): Likewise.
9549 (convert_move): Likewise.
9550 (alignment_for_piecewise_move): Likewise.
9551 (widest_int_mode_for_size): Likewise.
9552 (emit_block_move_via_movmem): Likewise.
9553 (copy_blkmode_to_reg): Likewise.
9554 (set_storage_via_setmem): Likewise.
9555 (compress_float_constant): Likewise.
9556 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9557 * optabs-query.c (get_best_extraction_insn): Likewise.
9558 * optabs.c (expand_binop): Likewise.
9559 (expand_twoval_unop): Likewise.
9560 (expand_twoval_binop): Likewise.
9561 (widen_leading): Likewise.
9562 (widen_bswap): Likewise.
9563 (expand_parity): Likewise.
9564 (expand_unop): Likewise.
9565 (prepare_cmp_insn): Likewise.
9566 (prepare_float_lib_cmp): Likewise.
9567 (expand_float): Likewise.
9568 (expand_fix): Likewise.
9569 (expand_sfix_optab): Likewise.
9570 * postreload.c (move2add_use_add2_insn): Likewise.
9571 * reg-stack.c (reg_to_stack): Likewise.
9572 * reginfo.c (choose_hard_reg_mode): Likewise.
9573 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9574 * stor-layout.c (mode_for_size): Likewise.
9575 (smallest_mode_for_size): Likewise.
9576 (mode_for_vector): Likewise.
9577 (finish_bitfield_representative): Likewise.
9578 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
9579 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
9580 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9581 * var-tracking.c (prepare_call_arguments): Likewise.
9583 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9584 Alan Hayward <alan.hayward@arm.com>
9585 David Sherwood <david.sherwood@arm.com>
9587 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
9588 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
9589 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
9590 * machmode.h (mode_size): Move earlier in file.
9591 (mode_precision): Likewise.
9592 (mode_inner): Likewise.
9593 (mode_nunits): Likewise.
9594 (mode_unit_size): Likewise.
9595 (unit_unit_precision): Likewise.
9596 (mode_wider): Likewise.
9597 (mode_2xwider): Likewise.
9598 (machine_mode): New class.
9599 (mode_to_bytes): New function.
9600 (mode_to_bits): Likewise.
9601 (mode_to_precision): Likewise.
9602 (mode_to_inner): Likewise.
9603 (mode_to_unit_size): Likewise.
9604 (mode_to_unit_precision): Likewise.
9605 (mode_to_nunits): Likewise.
9606 (GET_MODE_SIZE): Use mode_to_bytes.
9607 (GET_MODE_BITSIZE): Use mode_to_bits.
9608 (GET_MODE_PRECISION): Use mode_to_precision.
9609 (GET_MODE_INNER): Use mode_to_inner.
9610 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
9611 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
9612 (GET_MODE_NUNITS): Use mode_to_nunits.
9613 * system.h (ALWAYS_INLINE): New macro.
9614 * config/powerpcspe/powerpcspe-c.c
9615 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
9616 int for arg1_mode and arg2_mode.
9618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9619 Alan Hayward <alan.hayward@arm.com>
9620 David Sherwood <david.sherwood@arm.com>
9622 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9623 Prefix mode names with E_ in case statements.
9624 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9625 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
9626 (aarch64_split_simd_move): Likewise.
9627 (aarch64_gen_storewb_pair): Likewise.
9628 (aarch64_gen_loadwb_pair): Likewise.
9629 (aarch64_gen_store_pair): Likewise.
9630 (aarch64_gen_load_pair): Likewise.
9631 (aarch64_get_condition_code_1): Likewise.
9632 (aarch64_constant_pool_reload_icode): Likewise.
9633 (get_rsqrte_type): Likewise.
9634 (get_rsqrts_type): Likewise.
9635 (get_recpe_type): Likewise.
9636 (get_recps_type): Likewise.
9637 (aarch64_gimplify_va_arg_expr): Likewise.
9638 (aarch64_simd_container_mode): Likewise.
9639 (aarch64_emit_load_exclusive): Likewise.
9640 (aarch64_emit_store_exclusive): Likewise.
9641 (aarch64_expand_compare_and_swap): Likewise.
9642 (aarch64_gen_atomic_cas): Likewise.
9643 (aarch64_emit_bic): Likewise.
9644 (aarch64_emit_atomic_swap): Likewise.
9645 (aarch64_emit_atomic_load_op): Likewise.
9646 (aarch64_evpc_trn): Likewise.
9647 (aarch64_evpc_uzp): Likewise.
9648 (aarch64_evpc_zip): Likewise.
9649 (aarch64_evpc_ext): Likewise.
9650 (aarch64_evpc_rev): Likewise.
9651 (aarch64_evpc_dup): Likewise.
9652 (aarch64_gen_ccmp_first): Likewise.
9653 (aarch64_gen_ccmp_next): Likewise.
9654 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9655 (alpha_emit_xfloating_libcall): Likewise.
9656 (emit_insxl): Likewise.
9657 (alpha_arg_type): Likewise.
9658 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
9659 (arc_preferred_simd_mode): Likewise.
9660 (arc_secondary_reload): Likewise.
9661 (get_arc_condition_code): Likewise.
9662 (arc_print_operand): Likewise.
9663 (arc_legitimate_constant_p): Likewise.
9664 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9665 * config/arc/arc.md (casesi_load): Likewise.
9666 (casesi_compact_jump): Likewise.
9667 * config/arc/predicates.md (proper_comparison_operator): Likewise.
9668 (cc_use_register): Likewise.
9669 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9670 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9671 (arm_init_iwmmxt_builtins): Likewise.
9672 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
9673 (neon_expand_vector_init): Likewise.
9674 (arm_attr_length_move_neon): Likewise.
9675 (maybe_get_arm_condition_code): Likewise.
9676 (arm_emit_vector_const): Likewise.
9677 (arm_preferred_simd_mode): Likewise.
9678 (arm_output_iwmmxt_tinsr): Likewise.
9679 (thumb1_output_casesi): Likewise.
9680 (thumb2_output_casesi): Likewise.
9681 (arm_emit_load_exclusive): Likewise.
9682 (arm_emit_store_exclusive): Likewise.
9683 (arm_expand_compare_and_swap): Likewise.
9684 (arm_evpc_neon_vuzp): Likewise.
9685 (arm_evpc_neon_vzip): Likewise.
9686 (arm_evpc_neon_vrev): Likewise.
9687 (arm_evpc_neon_vtrn): Likewise.
9688 (arm_evpc_neon_vext): Likewise.
9689 (arm_validize_comparison): Likewise.
9690 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
9691 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
9692 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
9693 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
9694 (c6x_preferred_simd_mode): Likewise.
9695 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
9696 (epiphany_rtx_costs): Likewise.
9697 * config/epiphany/predicates.md (proper_comparison_operator):
9699 * config/frv/frv.c (condexec_memory_operand): Likewise.
9700 (frv_emit_move): Likewise.
9701 (output_move_single): Likewise.
9702 (output_condmove_single): Likewise.
9703 (frv_hard_regno_mode_ok): Likewise.
9704 (frv_matching_accg_mode): Likewise.
9705 * config/h8300/h8300.c (split_adds_subs): Likewise.
9706 (h8300_rtx_costs): Likewise.
9707 (h8300_print_operand): Likewise.
9708 (compute_mov_length): Likewise.
9709 (output_logical_op): Likewise.
9710 (compute_logical_op_length): Likewise.
9711 (compute_logical_op_cc): Likewise.
9712 (h8300_shift_needs_scratch_p): Likewise.
9713 (output_a_shift): Likewise.
9714 (compute_a_shift_length): Likewise.
9715 (compute_a_shift_cc): Likewise.
9716 (expand_a_rotate): Likewise.
9717 (output_a_rotate): Likewise.
9718 * config/i386/i386.c (classify_argument): Likewise.
9719 (function_arg_advance_32): Likewise.
9720 (function_arg_32): Likewise.
9721 (function_arg_64): Likewise.
9722 (function_value_64): Likewise.
9723 (ix86_gimplify_va_arg): Likewise.
9724 (ix86_legitimate_constant_p): Likewise.
9725 (put_condition_code): Likewise.
9726 (split_double_mode): Likewise.
9727 (ix86_avx256_split_vector_move_misalign): Likewise.
9728 (ix86_expand_vector_logical_operator): Likewise.
9729 (ix86_split_idivmod): Likewise.
9730 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
9731 (ix86_build_const_vector): Likewise.
9732 (ix86_build_signbit_mask): Likewise.
9733 (ix86_match_ccmode): Likewise.
9734 (ix86_cc_modes_compatible): Likewise.
9735 (ix86_expand_branch): Likewise.
9736 (ix86_expand_sse_cmp): Likewise.
9737 (ix86_expand_sse_movcc): Likewise.
9738 (ix86_expand_int_sse_cmp): Likewise.
9739 (ix86_expand_vec_perm_vpermi2): Likewise.
9740 (ix86_expand_vec_perm): Likewise.
9741 (ix86_expand_sse_unpack): Likewise.
9742 (ix86_expand_int_addcc): Likewise.
9743 (ix86_split_to_parts): Likewise.
9744 (ix86_vectorize_builtin_gather): Likewise.
9745 (ix86_vectorize_builtin_scatter): Likewise.
9746 (avx_vpermilp_parallel): Likewise.
9747 (inline_memory_move_cost): Likewise.
9748 (ix86_tieable_integer_mode_p): Likewise.
9749 (x86_maybe_negate_const_int): Likewise.
9750 (ix86_expand_vector_init_duplicate): Likewise.
9751 (ix86_expand_vector_init_one_nonzero): Likewise.
9752 (ix86_expand_vector_init_one_var): Likewise.
9753 (ix86_expand_vector_init_concat): Likewise.
9754 (ix86_expand_vector_init_interleave): Likewise.
9755 (ix86_expand_vector_init_general): Likewise.
9756 (ix86_expand_vector_set): Likewise.
9757 (ix86_expand_vector_extract): Likewise.
9758 (emit_reduc_half): Likewise.
9759 (ix86_emit_i387_round): Likewise.
9760 (ix86_mangle_type): Likewise.
9761 (ix86_expand_round_sse4): Likewise.
9762 (expand_vec_perm_blend): Likewise.
9763 (canonicalize_vector_int_perm): Likewise.
9764 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
9765 (expand_vec_perm_1): Likewise.
9766 (expand_vec_perm_interleave3): Likewise.
9767 (expand_vec_perm_even_odd_pack): Likewise.
9768 (expand_vec_perm_even_odd_1): Likewise.
9769 (expand_vec_perm_broadcast_1): Likewise.
9770 (ix86_vectorize_vec_perm_const_ok): Likewise.
9771 (ix86_expand_vecop_qihi): Likewise.
9772 (ix86_expand_mul_widen_hilo): Likewise.
9773 (ix86_expand_sse2_abs): Likewise.
9774 (ix86_expand_pextr): Likewise.
9775 (ix86_expand_pinsr): Likewise.
9776 (ix86_preferred_simd_mode): Likewise.
9777 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
9778 * config/i386/sse.md (*andnot<mode>3): Likewise.
9779 (<mask_codefor><code><mode>3<mask_name>): Likewise.
9780 (*<code><mode>3): Likewise.
9781 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
9782 (ia64_expand_atomic_op): Likewise.
9783 (ia64_arg_type): Likewise.
9784 (ia64_mode_to_int): Likewise.
9785 (ia64_scalar_mode_supported_p): Likewise.
9786 (ia64_vector_mode_supported_p): Likewise.
9787 (expand_vec_perm_broadcast): Likewise.
9788 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
9789 (iq2000_function_arg_advance): Likewise.
9790 (iq2000_function_arg): Likewise.
9791 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
9792 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
9793 (m68k_libcall_value): Likewise.
9794 (m68k_function_value): Likewise.
9795 (sched_attr_op_type): Likewise.
9796 * config/mcore/mcore.c (mcore_output_move): Likewise.
9797 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
9799 (microblaze_function_arg): Likewise.
9800 * config/mips/mips.c (mips16_build_call_stub): Likewise.
9801 (mips_print_operand): Likewise.
9802 (mips_mode_ok_for_mov_fmt_p): Likewise.
9803 (mips_vector_mode_supported_p): Likewise.
9804 (mips_preferred_simd_mode): Likewise.
9805 (mips_expand_vpc_loongson_even_odd): Likewise.
9806 (mips_expand_vec_unpack): Likewise.
9807 (mips_expand_vi_broadcast): Likewise.
9808 (mips_expand_vector_init): Likewise.
9809 (mips_expand_vec_reduc): Likewise.
9810 (mips_expand_msa_cmp): Likewise.
9811 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
9812 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
9813 (cc_flags_for_mode): Likewise.
9814 * config/msp430/msp430.c (msp430_print_operand): Likewise.
9815 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
9816 (nds32_output_casesi_pc_relative): Likewise.
9817 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9818 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
9819 (nvptx_gen_unpack): Likewise.
9820 (nvptx_gen_pack): Likewise.
9821 (nvptx_gen_shuffle): Likewise.
9822 (nvptx_gen_wcast): Likewise.
9823 (nvptx_preferred_simd_mode): Likewise.
9824 * config/pa/pa.c (pa_secondary_reload): Likewise.
9825 * config/pa/predicates.md (base14_operand): Likewise.
9826 * config/powerpcspe/powerpcspe-c.c
9827 (altivec_resolve_overloaded_builtin): Likewise.
9828 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
9830 (rs6000_preferred_simd_mode): Likewise.
9831 (output_vec_const_move): Likewise.
9832 (rs6000_expand_vector_extract): Likewise.
9833 (rs6000_split_vec_extract_var): Likewise.
9834 (reg_offset_addressing_ok_p): Likewise.
9835 (rs6000_legitimate_offset_address_p): Likewise.
9836 (rs6000_legitimize_address): Likewise.
9837 (rs6000_emit_set_const): Likewise.
9838 (rs6000_const_vec): Likewise.
9839 (rs6000_emit_move): Likewise.
9840 (spe_build_register_parallel): Likewise.
9841 (rs6000_darwin64_record_arg_recurse): Likewise.
9842 (swap_selector_for_mode): Likewise.
9843 (spe_init_builtins): Likewise.
9844 (paired_init_builtins): Likewise.
9845 (altivec_init_builtins): Likewise.
9846 (do_load_for_compare): Likewise.
9847 (rs6000_generate_compare): Likewise.
9848 (rs6000_expand_float128_convert): Likewise.
9849 (emit_load_locked): Likewise.
9850 (emit_store_conditional): Likewise.
9851 (rs6000_output_function_epilogue): Likewise.
9852 (rs6000_handle_altivec_attribute): Likewise.
9853 (rs6000_function_value): Likewise.
9854 (emit_fusion_gpr_load): Likewise.
9855 (emit_fusion_p9_load): Likewise.
9856 (emit_fusion_p9_store): Likewise.
9857 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
9858 (fusion_gpr_mem_load): Likewise.
9859 (fusion_addis_mem_combo_load): Likewise.
9860 (fusion_addis_mem_combo_store): Likewise.
9861 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
9862 (fusion_gpr_mem_load): Likewise.
9863 (fusion_addis_mem_combo_load): Likewise.
9864 (fusion_addis_mem_combo_store): Likewise.
9865 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9867 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
9868 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
9869 (rs6000_preferred_simd_mode): Likewise.
9870 (output_vec_const_move): Likewise.
9871 (rs6000_expand_vector_extract): Likewise.
9872 (rs6000_split_vec_extract_var): Likewise.
9873 (reg_offset_addressing_ok_p): Likewise.
9874 (rs6000_legitimate_offset_address_p): Likewise.
9875 (rs6000_legitimize_address): Likewise.
9876 (rs6000_emit_set_const): Likewise.
9877 (rs6000_const_vec): Likewise.
9878 (rs6000_emit_move): Likewise.
9879 (rs6000_darwin64_record_arg_recurse): Likewise.
9880 (swap_selector_for_mode): Likewise.
9881 (paired_init_builtins): Likewise.
9882 (altivec_init_builtins): Likewise.
9883 (rs6000_expand_float128_convert): Likewise.
9884 (emit_load_locked): Likewise.
9885 (emit_store_conditional): Likewise.
9886 (rs6000_output_function_epilogue): Likewise.
9887 (rs6000_handle_altivec_attribute): Likewise.
9888 (rs6000_function_value): Likewise.
9889 (emit_fusion_gpr_load): Likewise.
9890 (emit_fusion_p9_load): Likewise.
9891 (emit_fusion_p9_store): Likewise.
9892 * config/rx/rx.c (rx_gen_move_template): Likewise.
9893 (flags_from_mode): Likewise.
9894 * config/s390/predicates.md (s390_alc_comparison): Likewise.
9895 (s390_slb_comparison): Likewise.
9896 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
9897 (s390_vector_mode_supported_p): Likewise.
9898 (s390_cc_modes_compatible): Likewise.
9899 (s390_match_ccmode_set): Likewise.
9900 (s390_canonicalize_comparison): Likewise.
9901 (s390_emit_compare_and_swap): Likewise.
9902 (s390_branch_condition_mask): Likewise.
9903 (s390_rtx_costs): Likewise.
9904 (s390_secondary_reload): Likewise.
9905 (__SECONDARY_RELOAD_CASE): Likewise.
9906 (s390_expand_cs): Likewise.
9907 (s390_preferred_simd_mode): Likewise.
9908 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
9909 * config/sh/sh.c (sh_print_operand): Likewise.
9910 (dump_table): Likewise.
9911 (sh_secondary_reload): Likewise.
9912 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9913 * config/sh/sh.md (casesi_worker_1): Likewise.
9914 (casesi_worker_2): Likewise.
9915 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
9916 (fcc_comparison_operator): Likewise.
9917 * config/sparc/sparc.c (sparc_expand_move): Likewise.
9918 (emit_soft_tfmode_cvt): Likewise.
9919 (sparc_preferred_simd_mode): Likewise.
9920 (output_cbranch): Likewise.
9921 (sparc_print_operand): Likewise.
9922 (sparc_expand_vec_perm_bmask): Likewise.
9923 (vector_init_bshuffle): Likewise.
9924 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9925 (spu_vector_mode_supported_p): Likewise.
9926 (spu_expand_insv): Likewise.
9927 (spu_emit_branch_or_set): Likewise.
9928 (spu_handle_vector_attribute): Likewise.
9929 (spu_builtin_splats): Likewise.
9930 (spu_builtin_extract): Likewise.
9931 (spu_builtin_promote): Likewise.
9932 (spu_expand_sign_extend): Likewise.
9933 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9934 (tilegx_simd_int): Likewise.
9935 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
9936 (tilepro_simd_int): Likewise.
9937 * config/v850/v850.c (const_double_split): Likewise.
9938 (v850_print_operand): Likewise.
9939 (ep_memory_offset): Likewise.
9940 * config/vax/vax.c (vax_rtx_costs): Likewise.
9941 (vax_output_int_move): Likewise.
9942 (vax_output_int_add): Likewise.
9943 (vax_output_int_subtract): Likewise.
9944 * config/visium/predicates.md (visium_branch_operator): Likewise.
9945 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
9946 (visium_print_operand_address): Likewise.
9947 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9948 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
9949 (xtensa_expand_conditional_branch): Likewise.
9950 (xtensa_copy_incoming_a7): Likewise.
9951 (xtensa_output_literal): Likewise.
9952 * dfp.c (decimal_real_maxval): Likewise.
9953 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
9955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9956 Alan Hayward <alan.hayward@arm.com>
9957 David Sherwood <david.sherwood@arm.com>
9959 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
9960 (mode_nunits_inline): Likewise.
9961 (mode_inner_inline): Likewise.
9962 (mode_unit_size_inline): Likewise.
9963 (mode_unit_precision_inline): Likewise.
9964 (emit_insn_modes_h): Likewise. Also emit a #define of the
9966 (emit_mode_wider): Add an E_ prefix to mode names.
9967 (emit_mode_complex): Likewise.
9968 (emit_mode_inner): Likewise.
9969 (emit_mode_adjustments): Likewise.
9970 (emit_mode_int_n): Likewise.
9971 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
9972 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
9973 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
9974 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
9975 (CRC32_BUILTIN, ENTRY): Likewise.
9976 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
9977 (aarch64_pop_regs): Likewise.
9978 (aarch64_process_components): Likewise.
9979 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
9980 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
9981 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
9982 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
9983 * config/arm/arm.c (arm_init_libfuncs): Likewise.
9984 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
9986 * config/i386/i386-builtin.def (pcmpestr): Likewise.
9987 (pcmpistr): Likewise.
9988 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
9989 * config/mmix/mmix.c (mmix_output_condition): Likewise.
9990 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9992 * config/rl78/rl78.c (mduc_regs): Likewise.
9993 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9994 (htm_expand_builtin): Likewise.
9995 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
9996 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
9997 * config/xtensa/xtensa.c (print_operand): Likewise.
9998 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
9999 (NUM_MODE_VECTOR_INT): Likewise.
10000 * genoutput.c (null_operand): Likewise.
10001 (output_operand_data): Likewise.
10002 * genrecog.c (print_parameter_value): Likewise.
10003 * lra.c (debug_operand_data): Likewise.
10005 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10007 * dwarf2out.c (scompare_loc_descriptor_wide)
10008 (scompare_loc_descriptor_narrow): New functions, split out from...
10009 (scompare_loc_descriptor): ...here.
10010 * expmed.c (emit_store_flag_int): New function, split out from...
10011 (emit_store_flag): ...here.
10013 2017-08-30 Richard Biener <rguenther@suse.de>
10015 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
10016 (dwarf2out_early_finish): Move setting of AT_pubnames from
10017 early debug output to early finish.
10019 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10021 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
10022 and -mdata-region to the assembler.
10024 2017-08-30 Richard Biener <rguenther@suse.de>
10026 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
10028 (gen_subprogram_die): Add DW_AT_object_pointer only early.
10029 (dwarf2out_early_global_decl): Only generate a DIE for the
10030 abstract origin if it doesn't already exist or is a declaration DIE.
10031 (resolve_addr): Do not add the linkage name twice when
10032 generating a stub DIE for the DW_TAG_GNU_call_site target.
10034 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10036 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10037 Use machine_mode rather than int for arg1_mode.
10039 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
10042 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
10043 that the second argument of the built-in functions to unpack
10044 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
10045 switch statement instead a lot of if statements.
10046 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
10047 Allow 64-bit values to be in Altivec registers as well as
10048 traditional floating point registers.
10049 (pack<mode>, FMOVE128_VSX iterator): Likewise.
10051 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
10053 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
10054 MAX_REGS_PER_ADDRESS == 1.
10056 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
10058 * config/i386/i386.opt (flag_fentry): Do not init to -1.
10059 * config/i386/i386.c (ix86_option_override_internal): Simplify
10060 setting of opts->x_flag_entry.
10062 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10063 Jakub Jelinek <jakub@redhat.com>
10064 Richard Biener <rguenther@suse.de>
10066 PR tree-optimization/81503
10067 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
10068 folded constant fits in the target type; reorder tests for clarity.
10070 2017-08-29 Martin Liska <mliska@suse.cz>
10072 * passes.def: Include pass_lower_switch.
10073 * stmt.c (dump_case_nodes): Remove and move to
10074 tree-switch-conversion.
10075 (case_values_threshold): Likewise.
10076 (expand_switch_as_decision_tree_p): Likewise.
10077 (emit_case_decision_tree): Likewise.
10078 (expand_case): Likewise.
10079 (balance_case_nodes): Likewise.
10080 (node_has_low_bound): Likewise.
10081 (node_has_high_bound): Likewise.
10082 (node_is_bounded): Likewise.
10083 (emit_case_nodes): Likewise.
10084 (struct simple_case_node): New struct.
10085 (add_case_node): Remove.
10086 (emit_case_dispatch_table): Use vector instead of case_list.
10087 (reset_out_edges_aux): Remove.
10088 (compute_cases_per_edge): Likewise.
10089 (expand_case): Build list of simple_case_node.
10090 (expand_sjlj_dispatch_table): Use it.
10091 * tree-switch-conversion.c (struct case_node): Moved from
10092 stmt.c and adjusted.
10093 (emit_case_nodes): Likewise.
10094 (node_has_low_bound): Likewise.
10095 (node_has_high_bound): Likewise.
10096 (node_is_bounded): Likewise.
10097 (case_values_threshold): Likewise.
10098 (reset_out_edges_aux): Likewise.
10099 (compute_cases_per_edge): Likewise.
10100 (add_case_node): Likewise.
10101 (dump_case_nodes): Likewise.
10102 (balance_case_nodes): Likewise.
10103 (expand_switch_as_decision_tree_p): Likewise.
10104 (emit_jump): Likewise.
10105 (emit_case_decision_tree): Likewise.
10106 (try_switch_expansion): Likewise.
10107 (do_jump_if_equal): Likewise.
10108 (emit_cmp_and_jump_insns): Likewise.
10109 (fix_phi_operands_for_edge): New function.
10110 (record_phi_operand_mapping): Likewise.
10111 (class pass_lower_switch): New pass.
10112 (pass_lower_switch::execute): New function.
10113 (make_pass_lower_switch): Likewise.
10114 (conditional_probability):
10115 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
10116 * tree-pass.h: Add make_pass_lower_switch.
10118 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10121 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
10124 2017-08-29 Richard Biener <rguenther@suse.de>
10126 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
10127 we do not add a DW_AT_inline attribute twice.
10128 (gen_subprogram_die): Remove code setting DW_AT_inline on
10129 DECL_ABSTRACT_P nodes.
10131 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
10133 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
10134 calls to internal functions.
10135 (gimplify_modify_expr): Likewise.
10136 * tree-call-cdce.c (use_internal_fn): Likewise.
10137 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
10138 (convert_to_divmod): Set the nothrow flag.
10139 * tree-if-conv.c (predicate_mem_writes): Likewise.
10140 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10141 (vectorizable_call): Likewise.
10142 (vectorizable_store): Likewise.
10143 (vectorizable_load): Likewise.
10144 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
10145 (vect_recog_mask_conversion_pattern): Likewise.
10147 2017-08-29 Martin Liska <mliska@suse.cz>
10150 * gcc.c (driver_handle_option): Add new argument.
10151 * opts-common.c (handle_option): Pass
10152 target_option_override_hook.
10153 * opts-global.c (lang_handle_option): Add new option.
10154 (set_default_handlers): Add new argument.
10155 (decode_options): Likewise.
10156 * opts.c (target_handle_option): Likewise.
10157 (common_handle_option): Call target_option_override_hook.
10158 * opts.h (struct cl_option_handler_func): Add hook for
10159 target option override.
10160 (struct cl_option_handlers): Likewise.
10161 (set_default_handlers): Add new argument.
10162 (decode_options): Likewise.
10163 (common_handle_option): Likewise.
10164 (target_handle_option): Likewise.
10165 * toplev.c (toplev::main): Pass targetm.target_option.override
10168 2017-08-29 Richard Biener <rguenther@suse.de>
10169 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10171 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
10172 life to the active subtree.
10174 2017-08-28 Jeff Law <law@redhat.com>
10176 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
10177 derive_equivalences.
10178 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
10179 Code moved into....
10180 (edge_info::derive_equivalences): New private member function
10182 * tree-ssa-dom.c (class edge_info): Changed from a struct
10183 to a class. Add ctor/dtor, methods and data members.
10184 (edge_info::edge_info): Renamed from allocate_edge_info.
10185 Initialize additional members.
10186 (edge_info::~edge_info): New.
10187 (free_dom_edge_info): Delete the edge info.
10188 (record_edge_info): Use new class & associated member functions.
10189 Tighten forms for testing for edge equivalences.
10190 (record_temporary_equivalences): Iterate over the simple
10191 equivalences rather than assuming there's only one per edge.
10192 (cprop_into_successor_phis): Iterate over the simple
10193 equivalences rather than assuming there's only one per edge.
10194 (optimize_stmt): Use operand_equal_p rather than pointer
10195 equality for mini-DSE code.
10197 2017-08-28 Nathan Sidwell <nathan@acm.org>
10199 * gcc.c (execute): Fold SIGPIPE handling into switch
10200 statement. Adjust internal error message.
10202 2017-08-28 Richard Biener <rguenther@suse.de>
10205 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
10206 Do nothing for removed DIEs.
10208 2017-08-28 Richard Biener <rguenther@suse.de>
10210 PR tree-optimization/81977
10211 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
10214 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
10217 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
10218 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
10219 using targetm.gen_mem_thread_fence.
10221 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
10224 * config/i386/i386.md (*<btsc><mode>): Change operand 2
10225 predicate to register_operand. Reorder operands.
10226 (*btr<mode>): Ditto.
10227 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
10228 (*btr<mode>_mask): Ditto.
10230 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
10232 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
10233 * config/rs6000/xmmintrin.h: New file.
10234 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
10236 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10239 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
10240 parameter and_insn and return it.
10241 (recombine_lvx_pattern): Insert a copy to ensure availability of
10242 the base register of the copied masking operation at the point of
10243 the instruction replacement.
10244 (recombine_stvx_pattern): Likewise.
10246 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
10248 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
10249 undocumented switches.
10250 (-mpower9-dform-vector): Likewise.
10251 (-mpower9-dform): Likewise.
10252 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
10253 comments to delete references to -mpower9-dform* switches.
10254 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
10255 Delete reference to -mpower9-dform* switches, test for
10256 -mpower9-vector instead.
10257 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
10258 (OTHER_P9_VECTOR_MASKS): Likewise.
10259 (POWERPC_MASKS): Likewise.
10260 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
10261 tests against -mpower9-dform* to -mpower9-vector. Delete code
10262 that checked for -mpower9-dform* consistancy with other options.
10263 Add test for -mpower9-misc to enable other power9 switches.
10264 (rs6000_init_hard_regno_mode_ok): Likewise.
10265 (rs6000_option_override_internal): Likewise.
10266 (rs6000_emit_prologue): Likewise.
10267 (rs6000_emit_epilogue): Likewise.
10268 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
10269 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
10270 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
10272 (emit_fusion_p9_store): Likewise.
10273 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
10274 resetting these macros if the assembler does not support ISA 3.0
10276 (TARGET_P9_DFORM_VECTOR): Likewise.
10277 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
10278 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
10280 2017-08-25 Alan Modra <amodra@gmail.com>
10282 PR rtl-optimization/81747
10283 * cse.c (cse_extended_basic_block): Don't attempt to record
10284 equivalences for degenerate conditional branches that branch
10285 to their fall-through.
10287 2017-08-24 Martin Sebor <msebor@redhat.com>
10289 PR middle-end/81908
10290 * gimple-fold.c (size_must_be_zero_p): New function.
10291 (gimple_fold_builtin_memory_op): Call it.
10293 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
10295 * config/rs6000/mm_malloc.h: New file.
10297 2017-08-24 Bin Cheng <bin.cheng@arm.com>
10299 PR tree-optimization/81913
10300 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
10301 analysis when either IVs in condition can wrap.
10303 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
10305 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
10306 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
10308 2017-08-24 Richard Biener <rguenther@suse.de>
10311 * targhooks.c (default_target_can_inline_p): Properly
10312 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
10313 is present and always compare.
10314 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
10315 infer -mfpmath=sse from TARGET_SSE_P.
10316 (ix86_can_inline_p): Properly use target_option_default_node when
10317 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
10319 2017-08-24 Richard Biener <rguenther@suse.de>
10322 * dwarf2out.c (output_die): Handle flag_generate_offload like
10324 (output_comp_unit): Likewise.
10325 (gen_array_type_die): Likewise.
10326 (dwarf2out_early_finish): Likewise.
10327 (note_variable_value_in_expr): Likewise.
10328 (dwarf2out_finish): Likewise. Adjust assert.
10329 * cgraphunit.c (symbol_table::compile): Move setting of
10330 flag_generate_offload earlier ...
10331 (symbol_table::finalize_compilation_unit): ... here, before
10332 early debug finalization.
10334 2017-08-24 Richard Biener <rguenther@suse.de>
10336 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
10337 and ipa-fnsummary.h.
10338 (ix86_can_inline_p): When ix86_fpmath flags do not match
10339 check whether the callee uses FP math at all.
10341 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
10343 PR middle-end/81931
10344 * tree-ssanames.c (get_nonzero_bits): Use element_precision
10345 instead of TYPE_PRECISION.
10347 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
10348 Alan Hayward <alan.hayward@arm.com>
10349 David Sherwood <david.sherwood@arm.com>
10351 * combine.c (make_extraction): Use subreg_offset_from_lsb.
10353 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
10355 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
10357 (ix86_frame::stack_realign_allocate): New field.
10358 (struct machine_frame_state): Modify comments.
10359 (machine_frame_state::sp_realigned_fp_end): New field.
10360 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
10361 layout calculation.
10362 (sp_valid_at): Add assertion to assure no attempt to access invalid
10363 offset of a realigned stack.
10364 (fp_valid_at): Likewise.
10365 (choose_baseaddr): Modify comments.
10366 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
10367 ix86_expand_prologue.
10368 (ix86_expand_prologue): Modify stack realignment and allocation.
10369 (ix86_expand_epilogue): Modify comments.
10370 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
10371 avx2_runtime, avx512f, and avx512f_runtime.
10373 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
10375 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
10376 (mstackrealign): Do not init to -1.
10377 * config/i386/i386.c (ix86_option_override_internal):
10378 Check opts_set, not opts when setting default value of
10379 opts->x_ix86_force_align_arg_pointer.
10381 2017-08-23 Richard Biener <rguenther@suse.de>
10383 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
10384 lang_hooks.decl_printable_name.
10385 * print-rtl-function.c (print_rtx_function): Likewise.
10386 * tree-pretty-print.c (dump_function_header): Likewise.
10388 2017-08-23 Richard Biener <rguenther@suse.de>
10391 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
10392 -g0 at compile-time.
10394 2017-08-23 Tamar Christina <tamar.christina@arm.com>
10396 PR middle-end/19706
10397 * doc/sourcebuild.texi (Other hardware attributes):
10400 2017-08-23 Tamar Christina <tamar.christina@arm.com>
10402 PR middle-end/19706
10403 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
10404 Fix single-use check.
10406 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
10408 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
10410 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
10412 * doc/install.texi: Modify to add more details on running selected
10415 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
10417 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
10418 is combined with -mabi=ms.
10419 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
10422 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10424 PR tree-optimization/81488
10425 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
10426 and cached_basis fields.
10427 (MAX_SPREAD): New constant.
10428 (alloc_cand_and_find_basis): Initialize new fields.
10429 (clear_visited): New function.
10430 (create_phi_basis_1): Rename from create_phi_basis, set visited
10431 and cached_basis fields.
10432 (create_phi_basis): New wrapper function.
10433 (phi_add_costs_1): Rename from phi_add_costs, add spread
10434 parameter, set visited field, short-circuit when limits reached.
10435 (phi_add_costs): New wrapper function.
10436 (record_phi_increments_1): Rename from record_phi_increments, set
10438 (record_phi_increments): New wrapper function.
10439 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
10440 (phi_incr_cost): New wrapper function.
10441 (all_phi_incrs_profitable_1): Rename from
10442 all_phi_incrs_profitable, set visited field.
10443 (all_phi_incrs_profitable): New wrapper function.
10445 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10446 Alan Hayward <alan.hayward@arm.com>
10447 David Sherwood <david.sherwood@arm.com>
10449 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
10450 that takes the outer and inner modes.
10451 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
10452 comparison as the canonical test for a paradoxical subreg.
10453 * combine.c (simplify_set): Use paradoxical_subreg_p.
10454 (make_extraction): Likewise.
10455 (force_to_mode): Likewise.
10456 (rtx_equal_for_field_assignment_p): Likewise.
10457 (gen_lowpart_for_combine): Likewise.
10458 (simplify_comparison): Likewise.
10459 * cse.c (equiv_constant): Likewise.
10460 * expmed.c (store_bit_field_1): Likewise.
10461 * final.c (alter_subreg): Likewise.
10462 * fwprop.c (propagate_rtx): Likewise.
10463 (forward_propagate_subreg): Likewise.
10464 * ira-conflicts.c (ira_build_conflicts): Likewise.
10465 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
10466 * lra-constraints.c (curr_insn_transform): Likewise.
10467 (split_reg): Likewise.
10468 * lra-eliminations.c (move_plus_up): Likewise.
10469 (lra_eliminate_regs_1): Likewise.
10470 * recog.c (general_operand): Likewise.
10471 * ree.c (combine_reaching_defs): Likewise.
10472 * reload.c (push_reload): Likewise.
10473 (find_reloads): Likewise.
10474 * reload1.c (elimination_effects): Likewise.
10475 (compute_reload_subreg_offset): Likewise.
10476 (choose_reload_regs): Likewise.
10477 * rtlanal.c (subreg_lsb_1): Likewise.
10478 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10479 (simplify_subreg): Likewise.
10480 * var-tracking.c (track_loc_p): Likewise.
10481 * emit-rtl.c (byte_lowpart_offset): Likewise.
10482 (paradoxical_subreg_p): Delete out-of-line definition.
10484 2017-08-22 Jeff Law <law@redhat.com>
10486 PR tree-optimization/81741
10487 PR tree-optimization/71947
10488 * tree-ssa-dom.c: Include tree-inline.h.
10489 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
10490 equivalences if one is more expensive to compute than the other.
10491 * tree-ssa-scopedtables.h (class const_or_copies): Make
10492 record_const_or_copy_raw method private.
10493 (class avail_exprs_stack): New method simplify_binary_operation.
10494 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
10495 avail_exprs_stack::simplify_binary_operation as needed.
10496 (avail_exprs_stack::simplify_binary_operation): New function.
10498 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
10500 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
10501 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
10502 (DOT_SYMBOLS): Likewise.
10503 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
10504 (RELOCATABLE_NEEDS_FIXUP): Likewise.
10505 (RS6000_ABI_NAME): Likewise.
10506 (TARGET_CMODEL): Likewise.
10507 (TOC_SECTION_ASM_OP): Likewise.
10508 (SET_CMODEL): New macro.
10509 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
10511 2017-08-22 Richard Biener <rguenther@suse.de>
10513 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
10514 to allow for free-lang-data replacements similar to verify_type_variant.
10516 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10517 Alan Hayward <alan.hayward@arm.com>
10518 David Sherwood <david.sherwood@arm.com>
10520 * config/aarch64/aarch64.md (casesi): Use DImode rather than
10521 VOIDmode for the LABEL_REF.
10523 2017-08-22 Richard Biener <rguenther@suse.de>
10525 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
10527 2017-08-22 Richard Biener <rguenther@suse.de>
10529 * common.opt (feliminate-dwarf2-dups): Ignore.
10530 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
10531 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
10532 same_die_p_wrap, compute_section_prefix,
10533 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
10534 (comdat_symbol_id, comdat_symbol_number): Likewise.
10535 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
10537 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
10538 (output_die): Mark unreachable path unreachable.
10539 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
10540 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
10541 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
10542 (dwarf2out_early_finish): Likewise.
10544 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
10546 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
10548 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
10551 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
10552 not VAR_P. Filter attribute warnings with OPT_Wattributes.
10553 (avr_attribute_table) <io, io_low, address>: Initialize
10554 .decl_required with true.
10556 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
10558 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
10559 undocumented debugging options.
10560 (-mvsx-scalar-double): Likewise.
10561 (-mallow-df-permute): Likewise.
10562 (-mvectorize-builtins): Likewise.
10563 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10564 (rs6000_builtin_vectorized_function): Likewise.
10565 (rs6000_builtin_md_vectorized_function): Likewise.
10566 (rs6000_opt_vars): Likewise.
10568 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
10571 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
10572 (*btrq_imm): Rename from *btrq.
10573 (*btcq_imm): Rename from *btcq.
10574 (btsc): New code attribute.
10575 (*<btsc><mode>): New insn pattern.
10576 (*btr<mode>): Ditto.
10577 (*<btsc><mode>_mask): New insn_and_split pattern.
10578 (*btr<mode>_mask): Ditto.
10580 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10581 Alan Hayward <alan.hayward@arm.com>
10582 David Sherwood <david.sherwood@arm.com>
10584 * function.c (pad_below): Simplify padding calculation.
10586 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10587 Alan Hayward <alan.hayward@arm.com>
10588 David Sherwood <david.sherwood@arm.com>
10590 * target.def (function_prologue): Remove frame size argument.
10591 (function_epilogue): Likewise.
10592 * doc/tm.texi: Regenerate.
10593 * final.c (final_start_function): Update call to function_prologue.
10594 (final_end_function): Update call to function_epilogue.
10595 (default_function_pro_epilogue): Remove frame size argument.
10596 * output.h (default_function_pro_epilogue): Likewise.
10597 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
10598 (arm_output_function_prologue): Likewise.
10599 * config/frv/frv.c (frv_function_prologue): Likewise.
10600 (frv_function_epilogue): Likewise.
10601 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
10602 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
10603 (ia64_output_function_epilogue): Likewise.
10604 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
10605 (m32r_output_function_epilogue): Likewise.
10606 * config/microblaze/microblaze.c (microblaze_function_prologue)
10607 (microblaze_function_epilogue): Likewise.
10608 * config/mips/mips.c (mips_output_function_prologue): Likewise.
10609 (mips_output_function_epilogue): Likewise.
10610 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
10611 (mmix_target_asm_function_epilogue): Likewise.
10612 * config/msp430/msp430.c (msp430_start_function): Likewise.
10613 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
10614 (nds32_asm_function_epilogue): Likewise.
10615 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
10616 * config/pa/pa.c (pa_output_function_prologue): Likewise.
10617 (pa_output_function_epilogue): Likewise.
10618 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
10619 (rs6000_output_function_epilogue): Likewise.
10620 * config/rl78/rl78.c (rl78_start_function): Likewise.
10621 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
10622 (rs6000_output_function_epilogue): Likewise.
10623 * config/rx/rx.c (rx_output_function_prologue): Likewise.
10624 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
10625 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
10626 (sparc_asm_function_epilogue): Likewise.
10628 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10630 * tree.h (type_has_mode_precision_p): New function.
10631 * convert.c (convert_to_integer_1): Use it.
10632 * expr.c (expand_expr_real_2): Likewise.
10633 (expand_expr_real_1): Likewise.
10634 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
10635 * match.pd: Likewise.
10636 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
10637 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
10638 * tree-tailcall.c (process_assignment): Likewise.
10639 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10640 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
10641 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
10642 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10643 (vectorizable_assignment): Likewise.
10644 (vectorizable_shift): Likewise.
10645 (vectorizable_operation): Likewise.
10646 * tree-vrp.c (register_edge_assert_for_2): Likewise.
10648 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
10650 * match.pd: Add pow (C, x) simplification.
10652 2017-08-21 Richard Biener <rguenther@suse.de>
10654 PR tree-optimization/81900
10655 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
10656 for blocks with abnormal predecessors.
10657 (compute_antic): Do not set visited flag prematurely.
10659 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
10662 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
10664 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10666 * stor-layout.h (vector_type_mode): Move to...
10667 * tree.h (vector_type_mode): ...here.
10668 * stor-layout.c (vector_type_mode): Move to...
10669 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
10671 2017-08-21 Richard Biener <rguenther@suse.de>
10673 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
10674 register_external_die hooks.
10675 (debug_false_tree_charstarstar_uhwistar): Declare.
10676 (debug_nothing_tree_charstar_uhwi): Likewise.
10677 * debug.c (do_nothing_debug_hooks): Adjust.
10678 (debug_false_tree_charstarstar_uhwistar): New do nothing.
10679 (debug_nothing_tree_charstar_uhwi): Likewise.
10680 * dbxout.c (dbx_debug_hooks): Adjust.
10681 (xcoff_debug_hooks): Likewise.
10682 * sdbout.c (sdb_debug_hooks): Likewise.
10683 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10684 * dwarf2out.c (macinfo_label_base): New global.
10685 (dwarf2out_register_external_die): New function for the
10686 register_external_die hook.
10687 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
10688 (dwarf2_debug_hooks): Use them.
10689 (dwarf2_lineno_debug_hooks): Adjust.
10690 (struct die_struct): Add with_offset flag.
10691 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
10692 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
10693 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
10694 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
10695 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
10696 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
10697 defining section names for the early LTO debug variants.
10698 (reset_indirect_string): New helper.
10699 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
10700 (print_dw_val): Add support for offsetted symbol references.
10701 (get_ultimate_context): Split out from is_cxx.
10702 (is_cxx): Use get_ultimate_context.
10703 (is_fortran): Add decl overload.
10704 (compute_comp_unit_symbol): Split out worker from
10705 compute_section_prefix.
10706 (compute_section_prefix): Call compute_comp_unit_symbol and
10707 set comdat_type_p here.
10708 (output_die): Skip DIE symbol output for the LTO added one.
10709 Handle DIE symbol references with offset.
10710 (output_comp_unit): Guard section name mangling properly.
10711 For LTO debug sections emit a symbol at the section beginning
10712 which we use to refer to its DIEs.
10713 (add_abstract_origin_attribute): For DIEs registered via
10714 dwarf2out_register_external_die directly refer to the early
10715 DIE rather than indirectly through the shadow one we created.
10716 Remove obsolete call to dwarf2out_abstract_function for
10717 non-function/block origins.
10718 (gen_array_type_die): When generating early LTO debug do
10719 not emit DW_AT_string_length.
10720 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
10721 late when in LTO. As suggested place a gcc_unreachable for
10722 the DECL_ABSTRACT_P case.
10723 (gen_subprogram_die): Avoid another specification DIE
10724 for early built declarations/definitions for the late LTO case.
10725 (gen_variable_die): Add type references for late duplicated VLA dies
10727 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
10728 we have the abstract instance already.
10729 (process_scope_var): Adjust decl DIE contexts in LTO which
10730 first puts them in limbo.
10731 (gen_decl_die): Do not generate type DIEs late apart from
10732 types for VLAs or for decls we do not yet have a DIE. Do not
10733 call dwarf2out_abstract_function late.
10734 (dwarf2out_early_global_decl): Make sure to create DIEs
10735 for abstract instances of a decl first.
10736 (dwarf2out_late_global_decl): Adjust comment.
10737 (output_macinfo_op): With multiple macro sections use
10738 macinfo_label_base to distinguish labels.
10739 (output_macinfo): Likewise. Update macinfo_label_base.
10740 Pass in the line info label.
10741 (note_variable_value_in_expr): When generating LTO resolve
10742 all variable values here by generating DIEs as needed.
10743 (init_sections_and_labels): Add early LTO debug flag parameter
10744 and generate different sections and names if set. Add generation
10745 counter for the labels so we can have multiple of them.
10746 (reset_dies): Helper to allow DIEs to be output multiple times.
10747 (dwarf2out_finish): When outputting DIEs to the fat part of an
10748 LTO object first reset DIEs.
10749 (dwarf2out_early_finish): Output early DIEs when generating LTO.
10750 (modified_type_die): Check for decl_ultimate_origin being self
10752 (gen_type_die_with_usage): Likewise.
10753 (gen_typedef_die): Allow decl_ultimate_origin being self.
10754 (set_decl_abstract_flags): Remove.
10755 (set_block_abstract_flags): Likewise.
10756 (dwarf2out_abstract_function): Treat the early generated DIEs
10757 as the abstract copy and only add DW_AT_inline and
10758 DW_AT_artificial here and call set_decl_origin_self.
10759 If the DIE has an abstract origin don't do anything.
10760 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
10761 if we have none yet (Go fails to build one, PR78628).
10762 (variably_modified_type_p): Prevent endless recursion for Ada
10763 cyclic pointer types.
10764 * lto-streamer-in.c: Include debug.h.
10765 (dref_queue): New global.
10766 (lto_read_tree_1): Stream in DIE references.
10767 (lto_input_tree): Register DIE references.
10768 (input_function): Stream DECL_DEBUG_ARGS.
10769 * lto-streamer-out.c: Include debug.h.
10770 (lto_write_tree_1): Output DIE references.
10771 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
10772 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
10773 (output_function): Stream DECL_DEBUG_ARGS.
10774 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
10775 Stream DECL_ABSTRACT_ORIGIN.
10776 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
10777 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
10778 DECL_CONTEXT for file-scope decls.
10779 * lto-streamer.h (struct dref_entry): Declare.
10780 (dref_queue): Likewise.
10781 * cfgexpand.c (pass_expand::execute): Do not call the
10782 outlining_inline_function hook here.
10783 * lto-wrapper.c (debug_obj): New global.
10784 (tool_cleanup): Unlink it if required.
10785 (debug_objcopy): New function.
10786 (run_gcc): Handle early debug sections in the IL files by
10787 extracting them to separate files, partially linkin them and
10788 feeding the result back as result to the linker.
10789 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
10790 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
10791 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
10792 sections into a separate segment.
10793 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
10795 (darwin_asm_dwarf_section): Likewise.
10796 (darwin_asm_output_dwarf_offset): Likewise.
10797 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
10799 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10800 Alan Hayward <alan.hayward@arm.com>
10801 David Sherwood <david.sherwood@arm.com>
10803 * read-md.h (md_reader::record_potential_iterator_use): Replace
10804 pointer argument with an rtx and an index.
10805 * read-rtl.c (iterator_group::apply_iterator): Likewise.
10806 (apply_mode_iterator): Likewise.
10807 (apply_code_iterator): Likewise.
10808 (apply_int_iterator): Likewise.
10809 (apply_subst_iterator): Likewise.
10810 (record_iterator_use): Likewise.
10811 (record_attribute_use): Likewise.
10812 (md_reader::record_potential_iterator_use): Likewise. Update calls
10813 to record_iterator_use and apply_iterator.
10814 (iterator_use): Replace ptr with x and index.
10815 (attribute_use): Likewise.
10816 (apply_attribute_uses): Update calls to apply_iterator.
10817 (apply_iterators): Likewise. Update initialization of iterator_use.
10818 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
10819 and record_potential_iterator_use.
10820 (rtx_reader::read_rtx_operand): Likewise.
10822 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10823 Alan Hayward <alan.hayward@arm.com>
10824 David Sherwood <david.sherwood@arm.com>
10826 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
10829 2017-08-21 Richard Biener <rguenther@suse.de>
10831 PR middle-end/81884
10832 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
10833 at struct end conservatively when comparing common bases.
10835 2017-08-21 Richard Biener <rguenther@suse.de>
10837 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
10838 (mem_ref_in_stmt): Remove.
10839 (determine_max_movement): Use ref index to get at the reference.
10840 (invariantness_dom_walker::before_dom_children): Deal with
10841 lim data already initialized.
10842 (gather_mem_refs_stmt): Initialize lim data and record ref index.
10844 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10846 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
10847 (TARGET_ISA_ROUND): Ditto.
10848 (TARGET_ROUND): Ditto.
10849 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
10850 * config/i386/i386.md: Ditto.
10851 * config/i386/sse.md: Ditto.
10852 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
10853 with OPTION_MASK_ISA_SSE4_1.
10855 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10858 * doc/extend.texi (x86 Built-in Functions): Correct the name of
10859 __builtin_ia32_lzcnt_u16.
10861 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10864 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
10865 (rs6000_set_current_function): Rewrite function to use it.
10867 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
10870 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
10871 and TYPE_WARN_IF_NOT_ALIGN.
10872 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
10873 (handle_warn_if_not_align): New.
10874 (place_union_field): Call handle_warn_if_not_align.
10875 (place_field): Call handle_warn_if_not_align.
10876 Copy TYPE_WARN_IF_NOT_ALIGN.
10877 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
10878 (layout_type): Likewise.
10879 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
10881 (tree_decl_common): Add warn_if_not_align.
10882 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
10883 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
10884 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
10885 (DECL_WARN_IF_NOT_ALIGN): Likewise.
10886 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
10887 * doc/extend.texi: Document warn_if_not_aligned attribute.
10888 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
10890 2017-08-17 Martin Liska <mliska@suse.cz>
10893 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
10894 (get_data_dependence): Use it as pointer type.
10895 (distribute_loop): Likewise.
10897 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10899 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
10900 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
10901 (p8_vmrgow_<mode>_direct): New define_insn.
10902 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
10903 handle endianness for vmrgew and vmrgow permute patterns.
10905 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
10907 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
10908 * config/rs6000/rs6000-cpus.def: Remove comment.
10909 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
10910 (POWERPC_MASKS): Likewise.
10911 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
10912 use of TARGET_VSX_TIMODE.
10913 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
10914 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
10915 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
10916 (rs6000_option_override_internal): Remove dead code.
10917 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
10918 (rs6000_legitimize_reload_address): Likewise.
10919 (rs6000_legitimate_address_p): Likewise.
10920 (rs6000_opt_masks): Delete "vsx-timode".
10921 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
10922 from function comment.
10923 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
10924 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
10925 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
10927 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
10928 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
10929 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
10930 (VSX_TI): Likewise.
10932 (define_peephole2): Likewise.
10934 2017-08-17 Martin Sebor <msebor@redhat.com>
10937 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
10938 past the end of an array.
10939 (test_pp_format): Add test cases.
10941 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
10943 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
10944 missing ECF_NOTHROW flags.
10946 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
10949 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
10950 operands residing in integer registers.
10951 (*vsx_le_perm_load_<mode>): Likewise.
10952 (*vsx_le_perm_store_<mode>): Likewise.
10953 (define_peephole2): Add peepholes to optimize the above.
10955 2017-08-17 Marek Polacek <polacek@redhat.com>
10957 PR middle-end/81814
10958 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
10959 to mimic what shorten_compare did. Change the return type to bool.
10960 (fold_cond_expr_with_comparison): Update call to
10961 operand_equal_for_comparison_p.
10962 (fold_ternary_loc): Likewise.
10964 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
10966 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
10968 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
10970 2017-08-17 Richard Biener <rguenther@suse.de>
10972 * tree-ssa-structalias.c (solve_graph): When propagating
10973 to successors update the graphs succ edges and avoid duplicate work.
10975 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
10978 * config/i386/i386.c (ix86_option_override_internal): Save target
10979 specific options after ix86_stack_protector_guard_reg was changed.
10981 2017-08-17 Richard Biener <rguenther@suse.de>
10983 PR tree-optimization/81827
10984 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
10985 (new_var_info): Initialize it conservatively.
10986 (get_call_vi): Mark register vars.
10987 (new_scalar_tmp_constraint_exp): Likewise.
10988 (handle_rhs_call): Likewise.
10989 (handle_const_call): Likewise.
10990 (create_function_info_for): Likewise.
10991 (solve_constraints): Sort varinfos to separate register from
10992 non-register vars to pack points-to solution bitmaps during
10995 2017-08-17 Marek Polacek <polacek@redhat.com>
10997 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
10999 2017-08-17 Richard Biener <rguenther@suse.de>
11001 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
11002 to true when overflow is undefined and we saturated the result.
11004 2017-08-17 Alan Modra <amodra@gmail.com>
11007 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
11008 Don't use store multiple if only one reg needs saving.
11009 (interesting_frame_related_regno): New function.
11010 (rs6000_frame_related): Don't emit frame info for regs that
11012 (rs6000_emit_epilogue): Likewise.
11014 2017-08-16 Nathan Sidwell <nathan@acm.org>
11016 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
11017 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
11018 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
11019 * tree.c (free_lang_data_in_type): Use else-if chain. Always
11020 clear TYPE_LANG_1. Remove obsolete member-function stripping.
11021 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
11022 (verify_type): Adjust for TYPE_BINFO move.
11023 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
11024 process TYPE_BINFO directly.
11025 (hash_tree): Likewise.
11026 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11028 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11031 2017-08-16 David Malcolm <dmalcolm@redhat.com>
11033 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
11035 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
11038 * config/i386/i386.md (*anddi_1_btr): Change predicates of
11039 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
11040 Add ix86_binary_operator_ok to insn constraint.
11041 (*iordi_1_bts): Ditto.
11042 (*xordi_1_btc): Ditto.
11043 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
11044 Update corresponding peephole2 pattern.
11048 2017-08-16 Bin Cheng <bin.cheng@arm.com>
11050 PR tree-optimization/81832
11051 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
11052 copy loop header which has IFN_LOOP_DIST_ALIAS call.
11054 2017-08-16 Marek Polacek <polacek@redhat.com>
11057 * fold-const.c (fold_indirect_ref_1): Restore original behavior
11058 regarding size_zero_node.
11060 2017-08-16 Martin Liska <mliska@suse.cz>
11063 * config.gcc: Respect previously set extra_objs in case
11066 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
11068 PR tree-optimization/81835
11069 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
11070 the phi SSA_NAME. Check that the condition in a COND_EXPR does
11071 not depend on the phi.
11073 2017-08-16 Alan Modra <amodra@gmail.com>
11075 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
11078 2017-08-16 Alan Modra <amodra@gmail.com>
11080 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
11081 (save_reg_p): ..into this. Update all callers.
11082 (first_reg_to_save): Simplify.
11084 2017-08-16 Alan Modra <amodra@gmail.com>
11086 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
11089 2017-08-15 Joseph Myers <joseph@codesourcery.com>
11093 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
11094 constant count if that count is less than 32.
11096 2017-08-15 Nathan Sidwell <nathan@acm.org>
11098 * gcc.c (execute): Emit friendlier message if inferior is killed
11099 by an external cause.
11101 2017-08-15 Richard Biener <rguenther@suse.de>
11103 PR tree-optimization/81790
11104 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
11105 CONSTRUCTORs from simplifying and VN.
11107 2017-08-14 Martin Sebor <msebor@redhat.com>
11109 * builtin-attrs.def: Add comments.
11111 2017-08-14 Martin Sebor <msebor@redhat.com>
11114 * doc/extend.texi (attribute nonstring): Document new attribute.
11116 2017-08-14 Martin Sebor <msebor@redhat.com>
11119 * tree-diagnostic.c (default_tree_printer): Handle %G.
11120 * gimple-pretty-print.h (percent_G_format): Declare new function.
11121 * gimple-pretty-print.c (percent_G_format): Define.
11122 * tree-pretty-print.c (percent_K_format): Add argument.
11124 2017-08-14 Martin Sebor <msebor@redhat.com>
11126 PR translation/79998
11127 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
11128 Remove a stray space.
11130 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
11133 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
11134 (*iordi_1_bts): Ditto.
11135 (*xordi_1_btc): Ditto.
11137 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11140 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
11141 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11143 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
11144 quoted strings, and make more translator-friendly.
11145 (darwin_rs6000_override_options): Likewise.
11146 (rs6000_option_override_internal): Likewise.
11147 (rs6000_return_in_memory): Fix overlong line.
11148 (init_cmulative_args): Use quoted strings, and make more
11149 translator-friendly.
11150 (rs6000_pass_by_reference): Fix overlong line.
11151 (def_builtin): Use quoted strings.
11152 (altivec_expand_predicate_builtin): Use quoted strings, and make
11153 more translator-friendly.
11154 (htm_expand_builtin): Use quoted strings.
11155 (cpu_expand_builtin): Use quoted strings, and make more
11156 translator-friendly.
11157 (altivec_expand_builtin): Likewise.
11158 (paired_expand_predicate_builtin): Likewise.
11159 (rs6000_invalid_builtin): Likewise.
11160 (builtin_function_type): Use quoted strings.
11161 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
11162 more translator-friendly.
11163 (rs6000_trampoline_init): Likewise.
11164 (rs6000_handle_altivec_attribute): Likewise.
11165 (rs6000_inner_target_options): Use quoted strings.
11166 (rs6000_disable_incompatible_switches): Likewise.
11167 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
11168 strings, and make more translator-friendly.
11169 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
11171 2017-08-14 Bin Cheng <bin.cheng@arm.com>
11173 PR tree-optimization/81799
11174 * tree-loop-distribution.c (version_loop_by_alias_check): Force
11175 cond_expr to simple gimple operand.
11177 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
11179 PR middle-end/46932
11180 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
11182 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
11186 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
11187 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
11188 TARGET_GASISR_PROLOGUES.
11189 * config/avr/avr.c (avr_option_override): Same.
11190 (avr_pass_pre_proep::execute): Same.
11192 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
11195 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
11196 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
11198 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
11200 * config/i386/i386.md (*load_tp_<mode>): Redefine as
11201 define_insn_and_split. Split to a memory load from 0 in
11202 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
11203 using PTR mode iterator.
11204 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
11205 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
11206 (*add_tp_<mode>): Redefine as define_insn_and_split.
11207 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
11208 address space. Merge with *add_tp_x32 using PTR mode iterator.
11209 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
11210 Split to an add with a memory load from 0 in
11211 DEFAULT_TLS_SEG_REG address space.
11213 2017-08-12 Andrew Pinski <apinski@cavium.com>
11215 * config/aarch64/aarch64-option-extensions.def (rdma):
11216 Fix feature string to what Linux prints out in /proc/cpuinfo.
11218 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
11221 * dwarf2out.c (modified_type_die): For C typedef types that have
11222 an ultimate origin, process the ultimate origin instead of the
11224 (gen_typedef_die): Assert that input DECLs have no ultimate
11226 (gen_type_die_with_usage): For typedef variants that have an
11227 ultimate origin, just call gen_decl_die on the original DECL.
11228 (process_scope_var): Avoid creating DIEs for local typedefs and
11229 concrete static variables.
11231 2017-08-12 Alan Modra <amodra@gmail.com>
11235 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
11236 match gnu-user.h startfile.
11237 (ENDFILE_LINUX_SPEC): Similarly.
11239 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
11242 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
11244 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
11246 2017-08-11 Tamar Christina <tamar.christina@arm.com>
11247 * config/aarch64/aarch64.md (mov<mode>): Change.
11248 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11249 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
11250 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
11252 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
11254 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
11255 for storage order barriers.
11257 2017-08-11 Martin Liska <mliska@suse.cz>
11260 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
11261 variables of void type.
11263 2017-08-11 Martin Liska <mliska@suse.cz>
11265 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
11266 TARGET_SUPPORTS_ALIASES.
11267 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
11268 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
11269 (optimize_weakref): Likewise.
11270 * symtab.c (symtab_node::noninterposable_alias): Likewise.
11271 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
11272 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
11274 2017-08-11 Martin Liska <mliska@suse.cz>
11277 * config/i386/i386.c (make_resolver_func): Do complete
11278 refactoring of the function.
11280 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
11283 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
11284 * config/i386/i386.c (ix86_stack_protect_guard): Use
11285 ix86_stack_protect_guard_symbol_str to generate varible declaration.
11286 * doc/invoke.texi (x86 Options): Document
11287 -mstack-protector-guard-symbol= option.
11289 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
11291 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
11292 * config/i386/i386.c (ix86_split_stack_guard): New function.
11293 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
11294 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
11295 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
11296 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
11297 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
11298 (split_stack_space_check): Call ix86_split_stack_guard.
11300 2017-08-10 Martin Sebor <msebor@redhat.com>
11302 * print-tree.c (print_node): Print location using the established
11304 Replace spaces with colons.
11305 (debug_raw, debug): Ditto.
11307 2017-08-10 Martin Sebor <msebor@redhat.com>
11310 * pretty-print.c (pp_format): Correct the handling of %s precision.
11312 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
11315 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11317 (ix86_finalize_stack_frame_flags): This. Also clear
11318 frame_pointer_needed if -fno-omit-frame-pointer is used without
11320 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11321 with ix86_finalize_stack_frame_flags.
11322 (ix86_expand_epilogue): Likewise.
11323 (ix86_expand_split_stack_prologue): Likewise.
11324 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
11326 2017-08-10 Martin Liska <mliska@suse.cz>
11329 * c-attribs.c (handle_target_attribute):
11330 Report warning for an empty string argument of target attribute.
11332 2017-08-09 Jakub Jelinek <jakub@redhat.com>
11335 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
11337 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
11338 or DECL_NONLOCAL labels.
11339 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
11340 or DECL_NONLOCAL labels here.
11342 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
11344 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
11345 to indicate when early gimple folding has been disabled.
11346 (rs6000_gimple_fold_builtin): Add debug content.
11347 (rs6000_invalid_builtin): Fix whitespace.
11348 (rs6000_expand_builtin): Fix whitespace.
11349 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
11351 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
11354 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
11355 SAVE_MULTIPLE if not all the registers that saves, should be saved.
11357 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
11359 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
11360 (qdf24xx): Likewise.
11361 * config/aarch64/aarch64.md: Include falkor.md.
11362 * config/aarch64/falkor.md: New.
11364 2017-08-09 Marek Polacek <polacek@redhat.com>
11367 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
11368 * diagnostic.c (emit_diagnostic): Add a comment.
11369 (emit_diagnostic_valist): New function.
11371 2017-08-09 Marek Polacek <polacek@redhat.com>
11374 * input.c (make_location): New overload.
11375 * input.h (make_location): Declare.
11377 2017-08-08 Alan Modra <amodra@gmail.com>
11378 H.J. Lu <hongjiu.lu@intel.com>
11381 * gcc.c (NO_PIE_SPEC): Delete.
11382 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
11384 (LINK_PIE_SPEC): ..to here.
11385 (LINK_COMMAND_SPEC): Support -no-pie.
11386 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
11387 chain of crtbegin*.o selection, update for PIE_SPEC changes and
11389 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
11390 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
11391 (ENDFILE_CRTEND_SPEC): Similarly.
11393 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
11396 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
11397 (mstack-protector-guard-offset=): Ditto.
11398 * config/i386/i386.c (ix86_option_override): Handle
11399 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
11401 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
11402 ix86_stack_protect_guard_offset variables.
11403 (TARGET_STACK_PROTECT_GUARD): Always define.
11404 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
11405 and -mstack-protector-guard-offset= options.
11407 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11409 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
11410 boundary case for the last candidate.
11412 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11414 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
11415 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
11417 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11419 PR middle-end/19706
11420 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
11421 * config/aarch64/aarch64-builtins.c
11422 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
11423 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
11424 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
11426 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11427 Andrew Pinski <pinskia@gmail.com>
11429 PR middle-end/19706
11430 * internal-fn.def (XORSIGN): New.
11431 * optabs.def (xorsign_optab): New.
11432 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
11433 (convert_expand_mult_copysign): New.
11434 (pass_optimize_widening_mul::execute): Call
11435 convert_expand_mult_copysign.
11437 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11439 PR tree-optimization/81354
11440 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
11441 Insert on edges rather than explicitly creating landing pads.
11442 (analyze_candidates_and_replace): Commit edge inserts.
11444 2017-08-08 Richard Biener <rguenther@suse.de>
11446 PR middle-end/81719
11447 * tree-ssa-loop-niter.c: Include tree-dfa.h.
11448 (expand_simple_operations): Also look through ADDR_EXPRs with
11449 MEM_REF bases treating them as POINTER_PLUS_EXPR.
11451 2017-08-08 Richard Biener <rguenther@suse.de>
11453 PR tree-optimization/81723
11454 * tree-vect-slp.c (struct bst_traits): New hash traits.
11455 (bst_fail): New global.
11456 (vect_build_slp_tree_2): New worker, split out from ...
11457 (vect_build_slp_tree): ... this now wrapping it with using
11458 bst_fail set to cache SLP tree build fails. Properly handle
11460 (vect_analyze_slp_instance): Allocate and free bst_fail.
11462 2017-08-08 Martin Liska <mliska@suse.cz>
11465 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
11466 LABEL_DECLs that can be from a different function.
11468 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11470 PR tree-optimization/81744
11471 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
11472 loop's number of iterations.
11474 2017-08-08 Martin Liska <mliska@suse.cz>
11476 * asan.c: Include header files.
11477 * attribs.c (build_decl_attribute_variant): New function moved
11479 (build_type_attribute_qual_variant): Likewise.
11480 (cmp_attrib_identifiers): Likewise.
11481 (simple_cst_list_equal): Likewise.
11482 (omp_declare_simd_clauses_equal): Likewise.
11483 (attribute_value_equal): Likewise.
11484 (comp_type_attributes): Likewise.
11485 (build_type_attribute_variant): Likewise.
11486 (lookup_ident_attribute): Likewise.
11487 (remove_attribute): Likewise.
11488 (merge_attributes): Likewise.
11489 (merge_type_attributes): Likewise.
11490 (merge_decl_attributes): Likewise.
11491 (merge_dllimport_decl_attributes): Likewise.
11492 (handle_dll_attribute): Likewise.
11493 (attribute_list_equal): Likewise.
11494 (attribute_list_contained): Likewise.
11495 * attribs.h (lookup_attribute): New function moved from tree.[ch].
11496 (lookup_attribute_by_prefix): Likewise.
11497 * bb-reorder.c: Include header files.
11498 * builtins.c: Likewise.
11499 * calls.c: Likewise.
11500 * cfgexpand.c: Likewise.
11501 * cgraph.c: Likewise.
11502 * cgraphunit.c: Likewise.
11503 * convert.c: Likewise.
11504 * dwarf2out.c: Likewise.
11505 * final.c: Likewise.
11506 * fold-const.c: Likewise.
11507 * function.c: Likewise.
11508 * gimple-expr.c: Likewise.
11509 * gimple-fold.c: Likewise.
11510 * gimple-pretty-print.c: Likewise.
11511 * gimple.c: Likewise.
11512 * gimplify.c: Likewise.
11513 * hsa-common.c: Likewise.
11514 * hsa-gen.c: Likewise.
11515 * internal-fn.c: Likewise.
11516 * ipa-chkp.c: Likewise.
11517 * ipa-cp.c: Likewise.
11518 * ipa-devirt.c: Likewise.
11519 * ipa-fnsummary.c: Likewise.
11520 * ipa-inline.c: Likewise.
11521 * ipa-visibility.c: Likewise.
11523 * lto-cgraph.c: Likewise.
11524 * omp-expand.c: Likewise.
11525 * omp-general.c: Likewise.
11526 * omp-low.c: Likewise.
11527 * omp-offload.c: Likewise.
11528 * omp-simd-clone.c: Likewise.
11529 * opts-global.c: Likewise.
11530 * passes.c: Likewise.
11531 * predict.c: Likewise.
11532 * sancov.c: Likewise.
11533 * sanopt.c: Likewise.
11534 * symtab.c: Likewise.
11535 * toplev.c: Likewise.
11536 * trans-mem.c: Likewise.
11537 * tree-chkp.c: Likewise.
11538 * tree-eh.c: Likewise.
11539 * tree-into-ssa.c: Likewise.
11540 * tree-object-size.c: Likewise.
11541 * tree-parloops.c: Likewise.
11542 * tree-profile.c: Likewise.
11543 * tree-ssa-ccp.c: Likewise.
11544 * tree-ssa-live.c: Likewise.
11545 * tree-ssa-loop.c: Likewise.
11546 * tree-ssa-sccvn.c: Likewise.
11547 * tree-ssa-structalias.c: Likewise.
11548 * tree-ssa.c: Likewise.
11549 * tree-streamer-in.c: Likewise.
11550 * tree-vectorizer.c: Likewise.
11551 * tree-vrp.c: Likewise.
11552 * tsan.c: Likewise.
11553 * ubsan.c: Likewise.
11554 * varasm.c: Likewise.
11555 * varpool.c: Likewise.
11556 * tree.c: Remove functions moved to attribs.[ch].
11557 * tree.h: Likewise.
11558 * config/aarch64/aarch64.c: Add attrs.h header file.
11559 * config/alpha/alpha.c: Likewise.
11560 * config/arc/arc.c: Likewise.
11561 * config/arm/arm.c: Likewise.
11562 * config/avr/avr.c: Likewise.
11563 * config/bfin/bfin.c: Likewise.
11564 * config/c6x/c6x.c: Likewise.
11565 * config/cr16/cr16.c: Likewise.
11566 * config/cris/cris.c: Likewise.
11567 * config/darwin.c: Likewise.
11568 * config/epiphany/epiphany.c: Likewise.
11569 * config/fr30/fr30.c: Likewise.
11570 * config/frv/frv.c: Likewise.
11571 * config/ft32/ft32.c: Likewise.
11572 * config/h8300/h8300.c: Likewise.
11573 * config/i386/winnt.c: Likewise.
11574 * config/ia64/ia64.c: Likewise.
11575 * config/iq2000/iq2000.c: Likewise.
11576 * config/lm32/lm32.c: Likewise.
11577 * config/m32c/m32c.c: Likewise.
11578 * config/m32r/m32r.c: Likewise.
11579 * config/m68k/m68k.c: Likewise.
11580 * config/mcore/mcore.c: Likewise.
11581 * config/microblaze/microblaze.c: Likewise.
11582 * config/mips/mips.c: Likewise.
11583 * config/mmix/mmix.c: Likewise.
11584 * config/mn10300/mn10300.c: Likewise.
11585 * config/moxie/moxie.c: Likewise.
11586 * config/msp430/msp430.c: Likewise.
11587 * config/nds32/nds32-isr.c: Likewise.
11588 * config/nds32/nds32.c: Likewise.
11589 * config/nios2/nios2.c: Likewise.
11590 * config/nvptx/nvptx.c: Likewise.
11591 * config/pa/pa.c: Likewise.
11592 * config/pdp11/pdp11.c: Likewise.
11593 * config/powerpcspe/powerpcspe.c: Likewise.
11594 * config/riscv/riscv.c: Likewise.
11595 * config/rl78/rl78.c: Likewise.
11596 * config/rx/rx.c: Likewise.
11597 * config/s390/s390.c: Likewise.
11598 * config/sh/sh.c: Likewise.
11599 * config/sol2.c: Likewise.
11600 * config/sparc/sparc.c: Likewise.
11601 * config/spu/spu.c: Likewise.
11602 * config/stormy16/stormy16.c: Likewise.
11603 * config/tilegx/tilegx.c: Likewise.
11604 * config/tilepro/tilepro.c: Likewise.
11605 * config/v850/v850.c: Likewise.
11606 * config/vax/vax.c: Likewise.
11607 * config/visium/visium.c: Likewise.
11608 * config/xtensa/xtensa.c: Likewise.
11610 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11613 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
11614 constraints since the -mupper-regs-* switches have been
11616 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
11617 into a vector from a double word element that was extracted from
11618 another vector, and eliminate extra XXPERMDI instructions.
11619 (vsx_concat_<mode>_2): Likewise.
11620 (vsx_concat_<mode>_3): Likewise.
11621 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
11622 concat to allow optimizing inserts from previous extracts.
11624 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
11626 * config/i386/i386.c (ix86_stack_protect_guard): Generate
11627 memory reference to a SSP offset in TLS address space.
11628 (ix86_print_operand) <case '@'>: Remove.
11629 (ix86_print_operand_punct_valid_p): Remove '@' code.
11630 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
11631 UNSPEC_SP_TLS_TEST.
11632 (stack_tls_protect_set_<mode>): Remove.
11633 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
11634 (stack_tls_protect_test_<mode>): Remove.
11635 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
11637 2017-08-07 Olivier Hainque <hainque@adacore.com>
11640 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
11642 2017-08-07 Douglas Rupp <rupp@adacore.com>
11644 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
11645 variable was referenced as multidir in command.
11647 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11650 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
11653 2017-08-07 Martin Liska <mliska@suse.cz>
11655 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
11656 * config/rl78/rl78.c: Add include of attribs.h.
11657 * config/sh/sh.c: Likewise.
11658 * config/v850/v850.c: Likewise.
11660 2017-08-07 Tom de Vries <tom@codesourcery.com>
11662 PR middle-end/78266
11663 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
11665 2017-08-07 Martin Liska <mliska@suse.cz>
11667 * config/mips/mips.c: Include attribs.h.
11669 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
11672 * doc/invoke.texi: Document change in behvaior for -Ofast for
11675 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
11677 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
11679 (aarch64_pop_regs): Likewise.
11680 (aarch64_gen_load_pair): Likewise.
11681 (aarch64_save_callee_saves): Likewise.
11682 (aarch64_restore_callee_saves): Likewise.
11684 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11686 * config/i386/i386.c: Revert the last change.
11688 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11691 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11693 (ix86_finalize_stack_frame_flags): This. Also clear
11694 frame_pointer_needed if -fno-omit-frame-pointer is used without
11696 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11697 with ix86_finalize_stack_frame_flags.
11698 (ix86_expand_epilogue): Likewise.
11699 (ix86_expand_split_stack_prologue): Likewise.
11701 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11704 * config/i386/i386.c (get_builtin_code_for_version): Set priority
11705 to P_AES for Westmere.
11707 2017-08-07 Jonathan Yong <10walls@gmail.com>
11709 * config/i386/mingw.opt (fset-stack-executable): Removed.
11710 * config/i386/cygming.opt (fset-stack-executable): Moved
11712 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
11714 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
11716 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
11718 2017-08-07 Marek Polacek <polacek@redhat.com>
11720 PR middle-end/81737
11721 * fold-const.c (fold_indirect_ref_1): Check type_domain.
11723 2017-08-07 Martin Liska <mliska@suse.cz>
11725 * attribs.h (canonicalize_attr_name): New function.
11726 (cmp_attribs): Move from c-format.c and adjusted.
11727 (is_attribute_p): Moved from tree.h.
11728 * tree-inline.c: Add new includes.
11729 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
11730 (private_is_attribute_p): Remove.
11731 (private_lookup_attribute): Likewise.
11732 (private_lookup_attribute_by_prefix): Simplify.
11733 (remove_attribute): Use is_attribute_p.
11734 * tree.h: Remove removed declarations.
11736 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11738 PR middle-end/81698
11739 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
11740 instead of computing it in the function. Formatting fix.
11741 (expand_case): Don't rely on default_edge being the first edge,
11742 clear it if removing it, pass default_edge to
11743 emit_case_dispatch_table.
11744 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
11747 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
11749 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
11750 insn in the function, emit NOP after the insn.
11752 2017-08-06 Tom de Vries <tom@codesourcery.com>
11754 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
11757 2017-08-06 Tom de Vries <tom@codesourcery.com>
11759 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
11762 2017-08-04 Yury Gribov <tetra2005@gmail.com>
11764 PR tree-optimization/57371
11765 * match.pd: New pattern.
11767 2017-08-04 Marek Polacek <polacek@redhat.com>
11769 PR middle-end/81695
11770 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
11771 perform the computation in offset_int.
11773 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11775 PR tree-optimization/81136
11776 * tree-vectorizer.h: Include tree-hash-traits.h.
11777 (vec_base_alignments): New typedef.
11778 (vec_info): Add a base_alignments field.
11779 (vect_record_base_alignments): Declare.
11780 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
11782 (DR_IS_CONDITIONAL_IN_STMT): New macro.
11783 (create_data_ref): Add an is_conditional_in_stmt argument.
11784 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
11785 the is_conditional_in_stmt field.
11786 (data_ref_loc): Add an is_conditional_in_stmt field.
11787 (get_references_in_stmt): Set the is_conditional_in_stmt field.
11788 (find_data_references_in_stmt): Update call to create_data_ref.
11789 (graphite_find_data_references_in_stmt): Likewise.
11790 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
11791 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
11792 (vect_record_base_alignment): New function.
11793 (vect_record_base_alignments): Likewise.
11794 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
11795 for nested statements even if we fail to compute a misalignment.
11796 Use pooled base alignments for unconditional references.
11797 (vect_find_same_alignment_drs): Compare base addresses instead
11799 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
11800 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
11802 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11804 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
11805 Add an explicit name for the enum. Use auto_vec for slp_instances
11806 and grouped_stores.
11807 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
11809 (_bb_vec_info): Add a constructor and destructor.
11810 (vinfo_for_stmt): Return NULL for uids of -1 as well.
11811 (destroy_loop_vec_info): Delete.
11812 (vect_destroy_datarefs): Likewise.
11813 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
11814 (vec_info::vec_info): New function.
11815 (vec_info::~vec_info): Likewise.
11816 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
11817 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
11818 destroy_loop_vec_info.
11819 * tree-vect-loop.c (new_loop_vec_info): Replace with...
11820 (_loop_vec_info::_loop_vec_info): ...this.
11821 (destroy_loop_vec_info): Replace with...
11822 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
11823 the stmt_vec_infos. Leave handling of vec_info information to its
11824 destructor. Remove explicit vector releases.
11825 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
11826 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
11827 * tree-vect-slp.c (new_bb_vec_info): Replace with...
11828 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
11829 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
11830 (destroy_bb_vec_info): Replace with...
11831 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
11832 information to its destructor.
11833 (vect_slp_analyze_bb_1): Use new and delete instead of
11834 new_bb_vec_info and destroy_bb_vec_info.
11835 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
11838 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11840 * tree-data-ref.h (subscript): Add access_fn field.
11841 (data_dependence_relation): Add could_be_independent_p.
11842 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
11843 (same_access_functions): Move to tree-data-ref.c.
11844 * tree-data-ref.c (ref_contains_union_access_p): New function.
11845 (access_fn_component_p): Likewise.
11846 (access_fn_components_comparable_p): Likewise.
11847 (dr_analyze_indices): Add a reference to access_fn_component_p.
11848 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
11850 (constant_access_functions): Likewise.
11851 (add_other_self_distances): Likewise.
11852 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
11853 (initialize_data_dependence_relation): Use XCNEW and remove
11854 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
11855 of access functions that have the same type. Allow the
11856 subsequence to end with different bases in some circumstances.
11857 Record the chosen access functions in SUB_ACCESS_FN.
11858 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
11859 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
11860 (subscript_dependence_tester_1): Likewise dra and drb.
11861 (build_classic_dist_vector): Update calls accordingly.
11862 (subscript_dependence_tester): Likewise.
11863 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
11864 DDR_COULD_BE_INDEPENDENT_P.
11865 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
11866 comp_alias_ddrs instead of may_alias_ddrs.
11867 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
11869 (vect_analyze_data_ref_dependence): Use it if
11870 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
11871 distance vectors if that fails.
11872 (dependence_distance_ge_vf): New function.
11873 (vect_prune_runtime_alias_test_list): Use it. Don't clear
11874 LOOP_VINFO_MAY_ALIAS_DDRS.
11876 2017-08-04 Richard Biener <rguenther@suse.de>
11878 PR middle-end/81705
11879 * fold-const.c (fold_binary_loc): Properly restrict
11880 minus_var0 && minus_var1 case when associating undefined overflow
11883 2017-08-04 Tom de Vries <tom@codesourcery.com>
11885 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
11887 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11889 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11890 Don't start diagnostic messages with a capital letter.
11891 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11893 (rs6000_invalid_builtin): Likewise.
11894 (rs6000_trampoline_init): Likewise.
11896 2017-08-03 Jakub Jelinek <jakub@redhat.com>
11899 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
11900 after setting changeable df flags.
11902 2017-08-03 Richard Biener <rguenther@suse.de>
11904 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
11905 up if the use is in USE - X.
11907 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
11909 * toplev.c (dumpfile.h): New include.
11910 (internal_error_reentered): New static function. Use it...
11911 (internal_error_function): ...here to handle reentered internal_error.
11913 2017-08-03 Richard Biener <rguenther@suse.de>
11915 PR middle-end/81148
11916 * fold-const.c (split_tree): Add minus_var and minus_con
11917 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
11918 here but always use minus_*.
11919 (associate_trees): Assert we never associate with MINUS_EXPR
11920 and NULL first operand. Do not recurse for PLUS_EXPR operands
11921 when associating as MINUS_EXPR either.
11922 (fold_binary_loc): Track minus_var and minus_con.
11924 2017-08-03 Tom de Vries <tom@codesourcery.com>
11927 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
11928 ACCEL_COMPILER, apply finish_options on
11929 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
11931 2017-08-03 Tom de Vries <tom@codesourcery.com>
11934 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
11935 function_entry_patch_area_size > 0.
11937 2017-08-03 Jakub Jelinek <jakub@redhat.com>
11940 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
11941 instead of 10??LU, perform unit multiplication in wide_int,
11942 don't change alloc_object_size_limit if the limit is larger
11945 PR tree-optimization/81655
11946 PR tree-optimization/81588
11947 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
11948 the case when ranges[i].low and high are 1 for unsigned type with
11951 PR middle-end/81052
11952 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
11953 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
11955 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11957 * tree-vrp.h: Add include guard.
11959 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
11962 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
11963 (ud2): New insn pattern.
11964 * config/i386/i386.c (ix86_expand_epilogue):
11965 For naked functions, generate ud2 instead of trap insn.
11967 2017-08-02 Marek Polacek <polacek@redhat.com>
11970 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
11972 2017-08-02 Tom de Vries <tom@codesourcery.com>
11973 Cesar Philippidis <cesar@codesourcery.com>
11975 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
11976 Add missing edge probabilities.
11978 2017-08-02 Tamar Christina <tamar.christina@arm.com>
11980 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
11981 Correct endianness.
11983 2017-08-02 Jakub Jelinek <jakub@redhat.com>
11985 PR middle-end/79499
11986 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
11987 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
11988 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
11989 prologue_seq sequences - if any.
11991 2017-08-02 Richard Biener <rguenther@suse.de>
11993 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
11994 via vectors if supported, integer extracts via punning if supported
11995 or otherwise vector extracts.
11997 2017-08-02 Richard Biener <rguenther@suse.de>
11999 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
12001 (bitmap_insert_into_set): ... this.
12003 2017-08-02 Richard Biener <rguenther@suse.de>
12005 PR tree-optimization/81633
12007 2015-08-17 Alan Hayward <alan.hayward@arm.com>
12009 PR tree-optimization/71752
12010 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
12012 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
12014 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
12015 (machine_function::call_ms2sysv_pad_out): Remove field.
12016 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
12017 (ix86_compute_frame_layout): Likewise.
12019 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
12022 * config/i386/i386.c (ix86_set_func_type): Disallow naked
12023 attribute with interrupt attribute.
12025 2017-08-01 Andrew Pinski <apinski@cavium.com>
12027 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
12028 BIT_INSERT_EXPR's operand 1
12029 to see if the types precision matches.
12031 2017-08-01 Martin Liska <mliska@suse.cz>
12033 PR middle-end/70140
12034 * builtins.c (expand_builtin_memcpy_args): Remove.
12035 (expand_builtin_memcpy): Call newly added function
12036 expand_builtin_memory_copy_args.
12037 (expand_builtin_memcpy_with_bounds): Likewise.
12038 (expand_builtin_mempcpy): Remove last argument.
12039 (expand_builtin_mempcpy_with_bounds): Likewise.
12040 (expand_builtin_memory_copy_args): New function created from
12041 expand_builtin_mempcpy_args with small modifications.
12042 (expand_builtin_mempcpy_args): Remove.
12043 (expand_builtin_stpcpy): Remove unused argument.
12044 (expand_builtin): Likewise.
12045 (expand_builtin_with_bounds): Likewise.
12047 2017-08-01 Martin Liska <mliska@suse.cz>
12050 Make mempcpy more optimal (PR middle-end/70140).
12052 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12055 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12056 __builtin_vec_cmpne verify both arguments are compatible vectors
12057 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
12058 verify arg1_type is a pointer or array type. For __builtin_vec_st,
12059 move computation of aligned to after checking the argument types.
12063 * config/rs6000/vsx.md (vextract_fp_from_shorth,
12064 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
12067 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
12068 Doug Rupp <rupp@adacore.com>
12069 Olivier Hainque <hainque@adacore.com>
12071 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
12072 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
12074 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
12075 for TARGET_OS_CPP_BUILTIN.
12076 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
12077 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
12079 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
12080 passing required abi options to the assembler for EABI configurations.
12081 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
12082 of .text.hot and .text.unlikely sections for kernel modules when
12083 using ARM style exceptions.
12084 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
12085 options. Add EXTRA_CC1_SPEC.
12086 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
12088 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
12090 (ARM_TARGET2_DWARF_FORMAT): Define.
12091 * config/arm/t-vxworks: Adjust multilib control to removal of the
12092 Diab command line options.
12094 2017-08-01 Martin Liska <mliska@suse.cz>
12096 PR gcov-profile/81561
12097 * gcov.c (unblock): Make unblocking safe as we need to preserve
12098 index correspondence of blocks and block_lists.
12100 2017-08-01 Richard Biener <rguenther@suse.de>
12102 PR tree-optimization/81181
12103 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
12104 (compute_antic): ... end of iteration here.
12106 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
12108 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
12109 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
12110 (ftree-slp-vectorize): Likewise.
12111 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
12112 can no longer be set independent of flag_tree_loop_vectorize.
12113 * omp-general.c (emp_max_vf): Likewise.
12114 * opts.c (enable_fdo_optimizations): Remove references to
12115 flag_tree_vectorize, these are now implicit.
12116 (common_handle_option): Remove handling for OPT_ftree_vectorize,
12117 and leave it for the options machinery.
12119 2017-08-01 Martin Liska <mliska@suse.cz>
12121 PR middle-end/70140
12122 * builtins.c (expand_builtin_memcpy_args): Remove.
12123 (expand_builtin_memcpy): Call newly added function
12124 expand_builtin_memory_copy_args.
12125 (expand_builtin_memcpy_with_bounds): Likewise.
12126 (expand_builtin_mempcpy): Remove last argument.
12127 (expand_builtin_mempcpy_with_bounds): Likewise.
12128 (expand_builtin_memory_copy_args): New function created from
12129 expand_builtin_mempcpy_args with small modifications.
12130 (expand_builtin_mempcpy_args): Remove.
12131 (expand_builtin_stpcpy): Remove unused argument.
12132 (expand_builtin): Likewise.
12133 (expand_builtin_with_bounds): Likewise.
12135 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
12138 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
12139 print "ds:" only for immediates in generic address space.
12141 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
12144 * config/i386/i386.c (ix86_funciton_naked): New prototype.
12145 (ix86_function_ok_for_sibcall): Return false for naked functions.
12147 2017-08-01 Richard Biener <rguenther@suse.de>
12149 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
12150 (compute_antic): Seed worklist with exit block predecessors.
12151 * cfganal.c (dfs_find_deadend): For a cycle return the source
12152 of the edge closing it.
12154 2017-08-01 Tamar Christina <tamar.christina@arm.com>
12156 * config/aarch64/aarch64.c
12157 (aarch64_can_const_movi_rtx_p): Move 0 check.
12159 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12161 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
12162 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
12164 * match.pd: Ditto in address comparison pattern.
12166 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12168 PR tree-optimization/81627
12169 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
12170 closed ssa form for store-store chain.
12172 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12174 PR tree-optimization/81620
12175 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
12176 for store-store chain.
12178 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12180 PR tree-optimization/81588
12181 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
12182 ranges[i].in_p, invert comparison code ccode. For >/>=,
12183 swap rhs1 and rhs2 and comparison code unconditionally,
12184 for </<= don't do that. Don't swap rhs1/rhs2 again if
12185 ranges[i].in_p, instead invert comparison code ccode if
12186 opcode or oe->rank is BIT_IOR_EXPR.
12189 * optabs.def (vec_extract_optab, vec_init_optab): Change from
12190 a direct optab to conversion optab.
12191 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
12192 with GET_MODE_INNER as last argument instead of optab_handler.
12193 * expmed.c (extract_bit_field_1): Likewise. Use vector from
12194 vector extraction if possible and optab is available.
12195 * expr.c (store_constructor): Use convert_optab_handler instead
12196 of optab_handler. Use vector initialization from smaller
12197 vectors if possible and optab is available.
12198 * tree-vect-stmts.c (vectorizable_load): Likewise.
12199 * doc/md.texi (vec_extract, vec_init): Document that the optabs
12200 now have two modes.
12201 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
12202 of vec_init from half-sized vectors with the same element mode.
12203 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
12204 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
12205 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
12206 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
12207 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
12208 after mode in gen_vec_extract* calls.
12209 (vec_extract<mode>): Renamed to ...
12210 (vec_extract<mode><ssescalarmodelower>): ... this.
12211 (vec_extract<mode><ssehalfvecmodelower>): New expander.
12212 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
12213 element mode after mode in gen_vec_init* calls.
12214 (VEC_INIT_HALF_MODE): New mode iterator.
12215 (vec_init<mode>): Renamed to ...
12216 (vec_init<mode><ssescalarmodelower>): ... this.
12217 (vec_init<mode><ssehalfvecmodelower>): New expander.
12218 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
12219 (vec_extractv2sfsf): ... this.
12220 (vec_initv2sf): Renamed to ...
12221 (vec_initv2sfsf): ... this.
12222 (vec_extractv2si): Renamed to ...
12223 (vec_extractv2sisi): ... this.
12224 (vec_initv2si): Renamed to ...
12225 (vec_initv2sisi): ... this.
12226 (vec_extractv4hi): Renamed to ...
12227 (vec_extractv4hihi): ... this.
12228 (vec_initv4hi): Renamed to ...
12229 (vec_initv4hihi): ... this.
12230 (vec_extractv8qi): Renamed to ...
12231 (vec_extractv8qiqi): ... this.
12232 (vec_initv8qi): Renamed to ...
12233 (vec_initv8qiqi): ... this.
12234 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
12235 (vec_init<mode>): Renamed to ...
12236 (vec_init<mode><VEC_base_l>): ... this.
12237 (vec_extract<mode>): Renamed to ...
12238 (vec_extract<mode><VEC_base_l>): ... this.
12239 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
12240 (vec_initv2sfsf): ... this.
12241 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
12242 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
12243 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
12244 element mode after mode in gen_vec_init* calls.
12245 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
12246 (vec_init<mode><Vel>): ... this.
12247 (vec_extract<mode>): Renamed to ...
12248 (vec_extract<mode><Vel>): ... this.
12249 * config/aarch64/iterators.md (Vel): New mode attribute.
12250 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
12251 Add element mode after mode in gen_vec_extract* calls.
12252 * config/s390/vector.md (non_vec_l): New mode attribute.
12253 (vec_extract<mode>): Renamed to ...
12254 (vec_extract<mode><non_vec_l>): ... this.
12255 (vec_init<mode>): Renamed to ...
12256 (vec_init<mode><non_vec_l>): ... this.
12257 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
12258 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
12260 * config/arm/iterators.md (V_elem_l): New mode attribute.
12261 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
12262 (vec_extract<mode><V_elem_l>): ... this.
12263 (vec_extractv2di): Renamed to ...
12264 (vec_extractv2didi): ... this.
12265 (vec_init<mode>): Renamed to ...
12266 (vec_init<mode><V_elem_l>): ... this.
12267 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
12268 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
12269 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
12270 Add element mode after gen_vec_extract* calls.
12271 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
12272 (vec_init<mode><unitmode>): ... this.
12273 (vec_extract<mode>): Renamed to ...
12274 (vec_extract<mode><unitmode>): ... this.
12275 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
12276 (vec_init<mode><unitmode>): ... this.
12277 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
12278 (vec_initv2sfsf): ... this.
12279 (vec_extractv2sf): Renamed to ...
12280 (vec_extractv2sfsf): ... this.
12281 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
12282 Add element mode after gen_vec_extract* calls.
12283 * config/mips/mips.md (unitmode): New mode iterator.
12284 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
12285 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
12286 * config/spu/spu.md (inner_l): New mode attribute.
12287 (vec_init<mode>): Renamed to ...
12288 (vec_init<mode><inner_l>): ... this.
12289 (vec_extract<mode>): Renamed to ...
12290 (vec_extract<mode><inner_l>): ... this.
12291 * config/sparc/sparc.md (veltmode): New mode iterator.
12292 (vec_init<VMALL:mode>): Renamed to ...
12293 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
12294 * config/ia64/vect.md (vec_initv2si): Renamed to ...
12295 (vec_initv2sisi): ... this.
12296 (vec_initv2sf): Renamed to ...
12297 (vec_initv2sfsf): ... this.
12298 (vec_extractv2sf): Renamed to ...
12299 (vec_extractv2sfsf): ... this.
12300 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
12301 (vec_init<mode>): Renamed to ...
12302 (vec_init<mode><VEC_base_l>): ... this.
12303 (vec_extract<mode>): Renamed to ...
12304 (vec_extract<mode><VEC_base_l>): ... this.
12305 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
12306 (vec_initv2sfsf): ... this.
12307 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
12308 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
12309 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
12310 gen_vec_init* calls.
12312 2017-08-01 Richard Biener <rguenther@suse.de>
12314 PR tree-optimization/81297
12315 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
12316 TREE_OVERFLOW from INTEGER_CSTs.
12318 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
12320 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
12322 2017-07-31 Carl Love <cel@us.ibm.com>
12324 * config/rs6000/rs6000-c: Add support for built-in functions
12325 vector signed char vec_xl_be (signed long long, signed char *);
12326 vector unsigned char vec_xl_be (signed long long, unsigned char *);
12327 vector signed int vec_xl_be (signed long long, signed int *);
12328 vector unsigned int vec_xl_be (signed long long, unsigned int *);
12329 vector signed long long vec_xl_be (signed long long, signed long long *);
12330 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
12331 vector signed short vec_xl_be (signed long long, signed short *);
12332 vector unsigned short vec_xl_be (signed long long, unsigned short *);
12333 vector double vec_xl_be (signed long long, double *);
12334 vector float vec_xl_be (signed long long, float *);
12335 * config/rs6000/altivec.h (vec_xl_be): Add #define.
12336 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
12337 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
12339 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
12340 (altivec_expand_builtin): Add switch statement to call
12341 altivec_expand_xl_be for each builtin.
12342 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
12343 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
12344 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
12345 __builtin_vsx_le_be_v16qi.
12346 * doc/extend.texi: Update the built-in documentation file for the
12347 new built-in functions.
12349 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
12352 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
12354 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12356 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12358 * config.gcc: Add z14.
12359 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
12360 CPU model numbers for z13s and z14.
12361 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
12363 * config/s390/s390-opts.h (enum processor_type): Rename
12364 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
12365 * config/s390/s390.c (processor_table): Add field for CPU name to
12366 be passed to Binutils.
12367 (s390_asm_output_machine_for_arch): Use the new field in
12368 processor_table for Binutils.
12369 (s390_expand_builtin): Replace arch12 with z14.
12370 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
12371 (s390_get_sched_attrmask): Likewise.
12372 (s390_get_unit_mask): Likewise.
12373 * config/s390/s390.opt: Add z14 to processor_type enum.
12375 2017-07-31 Martin Jambor <mjambor@suse.cz>
12378 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
12379 regardless of optimization level.
12381 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
12382 Martin Liska <mliska@suse.cz>
12384 * predict.def: Remove old comment and adjust probability.
12385 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
12386 PREDICT statements.
12388 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
12391 * config/i386/i386.c (ix86_function_naked): New function.
12392 (ix86_can_use_return_insn_p): Return false for naked functions.
12393 (ix86_expand_prologue): Skip prologue for naked functions.
12394 (ix86_expand_epilogue): Skip epilogue for naked functions
12395 and emit trap instruction.
12396 (ix86_warn_func_return): New function.
12397 (ix86_attribute_table): Add "naked" attribute specification.
12398 (TARGET_WARN_FUNC_RETURN): Define.
12399 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
12401 2017-07-31 Martin Liska <mliska@suse.cz>
12403 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
12404 (dump_gimple_bb_header): Always dump BB info.
12405 (pp_cfg_jump): Do not append info about BB when dumping a jump.
12407 2017-07-31 Martin Liska <mliska@suse.cz>
12410 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
12411 also with current_function_decl non-null equality.
12413 2017-07-31 Jakub Jelinek <jakub@redhat.com>
12416 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
12417 change type to the element type, instead add eltype variable and
12418 use it where we are interested in the element type.
12420 PR tree-optimization/81603
12421 * ipa-polymorphic-call.c
12422 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
12423 offset arithmetic in offset_int, bail out if the resulting bit offset
12424 doesn't fit into shwi.
12426 2017-07-31 Martin Liska <mliska@suse.cz>
12428 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
12429 (gimplify_save_expr): Fix comment.
12431 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
12434 * config/i386/i386.c (ix86_function_arg): Update arguments for
12436 (ix86_compute_frame_layout): Set the initial stack offset to
12437 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
12438 INCOMING_FRAME_SP_OFFSET.
12439 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
12440 stack before exception handler returns.
12441 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
12442 the 'ERROR_CODE' for exception handler.
12444 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
12446 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
12447 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
12448 (ASM_OUTPUT_REG_POP): Ditto.
12449 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
12450 instead of asm_fprintf to output pure string.
12452 2017-07-29 Jakub Jelinek <jakub@redhat.com>
12454 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
12455 to imported_module_or_decl hook.
12456 (debug_nothing_tree_tree_tree_bool): Remove.
12457 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
12458 * debug.c (do_nothing_debug_hooks): Use
12459 debug_nothing_tree_tree_tree_bool_bool instead of
12460 debug_nothing_tree_tree_tree_bool.
12461 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12462 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
12463 * sdbout.c (sdb_debug_hooks): Likewise.
12464 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
12465 (gen_namespace_die): Add DW_AT_export_symbols attribute if
12467 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
12468 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
12469 attribute, don't add anything.
12471 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12473 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
12474 (fold_build2_stat_loc): Likewise.
12475 (fold_build3_stat_loc): Likewise.
12476 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
12477 (fold_build1_loc): Remove macro.
12478 (fold_build2_loc): Likewise.
12479 (fold_build3_loc): Likewise.
12481 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12483 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
12484 (gimple_build_debug_bind_source_stat): Likewise.
12485 * gimple.h (gimple_build_debug_bind): Remove macro.
12486 (gimple_build_debug_bind_source): Likewise.
12488 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12490 * bitmap.c (bitmap_alloc): Adjust.
12491 (bitmap_gc_alloc): Likewise.
12492 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
12494 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12496 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
12497 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
12498 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
12499 (bitmap_gc_alloc_stat): Likewise.
12500 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
12502 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12504 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
12505 * rtl.h (shallow_copy_rtx): Remove macro.
12507 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12509 * emit-rtl.c (gen_raw_REG): Adjust.
12510 * gengenrtl.c (gendef): Likewise.
12511 * rtl.c (rtx_alloc_stat): Remove _stat from name.
12512 * rtl.h (rtx_alloc): Remove macro.
12514 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12516 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
12517 (build_tree_list_stat): Likewise.
12518 * tree.h (build_tree_list): Remove macro.
12519 (build_tree_list_vec): Likewise.
12521 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12523 * tree.c (make_vector_stat): Remove _stat from name.
12524 (build_vector_stat): Likewise.
12525 * tree.h (make_vector_stat): Remove macro.
12526 (build_vector_stat): Likewise.
12528 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12530 * tree.h (build_var_debug_value): Remove prototype.
12532 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12534 * tree.c (tree_cons_stat): Remove _stat from name.
12535 * tree.h (tree_cons): Remove macro.
12537 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12539 * tree.c (build_vl_exp_stat): Remove _stat from name.
12540 * tree.h (build_vl_exp): Remove macro.
12542 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12544 * tree.c (build_decl_stat): Remove _stat from name.
12545 * tree.h (build_decl): Remove macro.
12547 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12549 * gimple.c (gimple_build_with_ops_stat): Adjust.
12550 (gimple_alloc_stat): Remove _stat from name.
12551 * gimple.h (gimple_alloc): Remove macro.
12553 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12555 * tree.c (make_tree_vec_stat): Remove _stat from name.
12556 (grow_tree_vec_stat): Likewise.
12557 * tree.h (make_tree_vec_stat): Adjust prototype.
12558 (grow_tree_vec_stat): Likewise.
12559 (make_tree_vec): Remove macro.
12560 (grow_tree_vec): Likewise.
12562 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12564 * fold-const.c (fold_build1_stat_loc): Adjust.
12565 (fold_build2_stat_loc): Likewise.
12566 (fold_build3_stat_loc): Likewise.
12567 * tree.c (build0_stat): Remove _stat from name.
12568 (build1_stat): Likewise.
12569 (build2_stat): Likewise.
12570 (build3_stat): Likewise.
12571 (build4_stat): Likewise.
12572 (build5_stat): Likewise.
12573 * tree.h (build1_loc): Remove macro, and rename _stat function
12575 (build2_loc): Likewise.
12576 (build3_loc): Likewise.
12577 (build4_loc): Likewise.
12578 (build5_loc): Likewise.
12580 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12582 * tree.c (make_int_cst_stat): Remove _stat from name.
12583 * tree.h (make_int_cst_stat): Adjust prototype.
12584 (make_int_cst): Remove macro.
12586 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12588 * tree.c (make_tre_binfo_stat): Remove _stat from name.
12589 * tree.h (make_tree_binfo_stat): Adjust prototype.
12590 (make_tree_binfo): Remove.
12592 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12594 * tree.c (copy_node_stat): Rename to copy_node.
12595 (build_distinct_type_copy): Adjust.
12596 * tree.h (copy_node_stat): Adjust prototype.
12597 (copy_node): Remove macro.
12599 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12601 * tree.c (make_node_stat): rename to make_node.
12602 (build_tree_list_stat): Adjust.
12603 (build0_stat): Likewise.
12604 (build2_stat): Likewise.
12605 (build3_stat): Likewise.
12606 (build4_stat): Likewise.
12607 (build5_stat): Likewise.
12608 (build_decl_stat): Likewise.
12609 * tree.h (make_node_stat): Adjust prototype.
12610 (make_node): remove macro.
12612 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
12614 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
12615 (PPC_FEATURE2_SCV): Likewise.
12616 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
12618 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12620 * config/aarch64/aarch64.c
12621 (aarch64_internal_mov_immediate): Add new special pattern.
12622 * config/aarch64/aarch64.md (*movdi_aarch64):
12623 Add reg/32bit const mov case.
12625 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12626 Richard Sandiford <richard.sandiford@linaro.org>
12628 * config/aarch64/aarch64.md (mov<mode>): Generalize.
12629 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12630 Add integer and movi cases.
12631 (movi-split-hf-df-sf split, fp16): New.
12632 (enabled): Added TARGET_FP_F16INST.
12633 * config/aarch64/iterators.md (GPF_HF): New.
12634 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
12636 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12638 * config/aarch64/aarch64.c
12639 (aarch64_simd_container_mode): Add prototype.
12640 (aarch64_expand_mov_immediate): Add HI support.
12641 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
12642 (aarch64_can_const_movi_rtx_p): New.
12643 (aarch64_preferred_reload_class):
12644 Remove restrictions of using FP registers for certain SIMD operations.
12645 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
12646 (aarch64_valid_floating_const): Add integer move validation.
12647 (aarch64_simd_imm_scalar_p): Remove.
12648 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
12649 (aarch64_legitimate_constant_p): Expand list of supported cases.
12650 * config/aarch64/aarch64-protos.h
12651 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
12652 (aarch64_reinterpret_float_as_int): New.
12653 (aarch64_simd_imm_scalar_p): Remove.
12654 * config/aarch64/constraints.md (Uvi): New.
12655 (Dd): Split into Ds and new Dd.
12656 * config/aarch64/aarch64.md (*movsi_aarch64):
12658 (*movdi_aarch64): Add SIMD mov case.
12660 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12662 * tree-predcom.c: (struct chain): Handle store-store chain in which
12663 stores for elimination only store loop invariant values.
12664 (execute_pred_commoning_chain): Ditto.
12665 (prepare_initializers_chain_store_elim): Ditto.
12666 (prepare_finalizers): Ditto.
12667 (is_inv_store_elimination_chain): New function.
12668 (initialize_root_vars_store_elim_1): New function.
12670 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12672 * tree-predcom.c: Revise general description of the pass.
12673 (enum chain_type): New enum type for store elimination.
12674 (struct chain): New field supporting store elimination.
12675 (struct component): Ditto.
12676 (dump_chain): Dump store-stores chain.
12677 (release_chain): Release resources.
12678 (split_data_refs_to_components): Compute and create component
12679 contains only stores for elimination.
12680 (get_chain_last_ref_at): New function.
12681 (make_invariant_chain): Initialization.
12682 (make_rooted_chain): Specify chain type in parameter and record it.
12683 (add_looparound_copies): Skip for store-stores chain.
12684 (determine_roots_comp): Compute type of chain and pass it to
12686 (initialize_root_vars_store_elim_2): New function.
12687 (finalize_eliminated_stores): New function.
12688 (remove_stmt): Handle store for elimination.
12689 (execute_pred_commoning_chain): Execute predictive commoning on
12690 store-store chains.
12691 (determine_unroll_factor): Skip unroll for store-stores chain.
12692 (prepare_initializers_chain_store_elim): New function.
12693 (prepare_initializers_chain): Hanlde store-store chain.
12694 (prepare_finalizers_chain, prepare_finalizers): New function.
12695 (tree_predictive_commoning_loop): Return integer value indicating
12696 if loop is unrolled or lcssa form is corrupted.
12697 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
12699 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12701 * tree-predcom.c (initialize_root): Delete.
12702 (execute_pred_commoning_chain): Initialize root vars and replace
12703 reference of non-combined chain directly, rather than call above
12706 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12708 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
12709 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
12711 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12713 * tree-predcom.c (struct chain): New field init_seq.
12714 (release_chain): Release init_seq.
12715 (prepare_initializers_chain): Record intialization stmts in above
12717 (insert_init_seqs): New function.
12718 (tree_predictive_commoning_loop): Call insert_init_seqs.
12720 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12722 * tree-predcom.c (determine_roots_comp): Skip trivial components.
12724 2017-07-28 Richard Biener <rguenther@suse.de>
12726 * match.pd: Remove superfluous :c.
12727 * genmatch.c (simplify::id): Add member.
12728 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
12730 (current_id): New global.
12731 (dt_node::parent): Move from ...
12732 (dt_operand::parent): ... here. Add for_id member.
12733 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
12734 (decision_tree::find_node): Relax order requirement when
12735 merging DT_TRUE nodes to ones inbetween the current simplify
12736 and the one we try to merge with. Add diagnostic whenever
12737 we need to enforce pattern order by not merging.
12738 (decision_tree::insert): Set current_id.
12739 (decision_tree::print_node): Dump parent node and for_id.
12740 (parser::last_id): Add member.
12741 (parser::push_simplify): Assign unique id.
12742 (parser::parser): Initialize last_id.
12744 2017-07-28 Martin Liska <mliska@suse.cz>
12747 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
12748 gimple_build_debug_bind.
12750 2017-07-28 Richard Biener <rguenther@suse.de>
12752 PR tree-optimization/81502
12753 * match.pd: Add pattern combining BIT_INSERT_EXPR with
12755 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
12757 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
12758 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
12759 for BIT_FIELD_REF args.
12760 * fold-const.c (make_bit_field_ref): Likewise.
12761 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
12763 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12766 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
12767 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
12768 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
12769 Or it into SANITIZER_UNDEFINED.
12770 * ubsan.c: Include gimple-fold.h and varasm.h.
12771 (ubsan_expand_ptr_ifn): New function.
12772 (instrument_pointer_overflow): New function.
12773 (maybe_instrument_pointer_overflow): New function.
12774 (instrument_object_size): Formatting fix.
12775 (pass_ubsan::execute): Call instrument_pointer_overflow
12776 and maybe_instrument_pointer_overflow.
12777 * internal-fn.c (expand_UBSAN_PTR): New function.
12778 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
12779 * sanitizer.def (__ubsan_handle_pointer_overflow,
12780 __ubsan_handle_pointer_overflow_abort): New builtins.
12781 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
12782 * internal-fn.def (UBSAN_PTR): New internal function.
12783 * opts.c (sanitizer_opts): Add pointer-overflow.
12784 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
12785 * fold-const.c (build_range_check): Compute pointer range check in
12786 integral type if pointer arithmetics would be needed. Formatting
12789 2017-07-28 Martin Liska <mliska@suse.cz>
12792 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
12793 parameters that are of a variable-length.
12795 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
12797 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
12799 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
12800 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
12801 (CRT_CALL_STATIC_FUNCTION): Likewise.
12802 (ASM_DEFAULT_SPEC): New define.
12803 (ASM_SPEC32): Likewise.
12804 (ASM_SPEC64): Likewise.
12805 (ASM_SPEC_COMMON): Likewise.
12806 (ASM_SPEC): Likewise.
12807 (INVALID_64BIT): Likewise.
12808 (LINK_OS_DEFAULT_SPEC): Likewise.
12809 (LINK_OS_SPEC32): Likewise.
12810 (LINK_OS_SPEC64): Likewise.
12811 (POWERPC_LINUX): Likewise.
12812 (PTRDIFF_TYPE): Likewise.
12813 (RESTORE_FP_PREFIX): Likewise.
12814 (RESTORE_FP_SUFFIX): Likewise.
12815 (SAVE_FP_PREFIX): Likewise.
12816 (SAVE_FP_SUFFIX): Likewise.
12817 (SIZE_TYPE): Likewise.
12818 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
12819 (TARGET_64BIT): Likewise.
12820 (TARGET_64BIT): Likewise.
12821 (TARGET_AIX): Likewise.
12822 (WCHAR_TYPE_SIZE): Likewise.
12823 (WCHAR_TYPE): Undefine.
12824 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
12825 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
12826 (CPP_OS_RTEMS_SPEC): Delete.
12827 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
12828 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
12830 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
12832 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12834 PR tree-optimization/81578
12835 * tree-parloops.c (build_new_reduction): Bail out if
12836 reduction_code isn't one of the standard OpenMP reductions.
12837 Move the details printing after that decision.
12839 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12841 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
12842 related to reload_in_progress.
12843 (splat_input_operand): Likewise.
12844 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
12846 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
12848 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
12849 (TARGET_INSTANTIATE_DECLS): Likewise.
12850 (legitimate_indexed_address_p): Delete reload_in_progress code.
12851 (rs6000_debug_legitimate_address_p): Likewise.
12852 (rs6000_eliminate_indexed_memrefs): Likewise.
12853 (rs6000_emit_le_vsx_store): Likewise.
12854 (rs6000_emit_move_si_sf_subreg): Likewise.
12855 (rs6000_emit_move): Likewise.
12856 (register_to_reg_type): Likewise.
12857 (rs6000_pre_atomic_barrier): Likewise.
12858 (rs6000_machopic_legitimize_pic_address): Likewise.
12859 (rs6000_allocate_stack_temp): Likewise.
12860 (rs6000_address_for_fpconvert): Likewise.
12861 (rs6000_address_for_altivec): Likewise.
12862 (rs6000_secondary_memory_needed_rtx): Delete function.
12863 (rs6000_check_sdmode): Likewise.
12864 (rs6000_alloc_sdmode_stack_slot): Likewise.
12865 (rs6000_instantiate_decls): Likewise.
12866 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
12867 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
12868 Delete reload_in_progress.
12869 (*vec_reload_and_plus_<mptrsize>): Likewise.
12870 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
12871 (vsx_div_v2di): Likewise.
12872 (vsx_udiv_v2di): Likewise.
12874 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12876 * config/rs6000/rs6000.opt (mlra): Replace with stub.
12877 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
12878 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
12879 (rs6000_debug_reg_global): Delete print of LRA status.
12880 (rs6000_option_override_internal): Delete dead LRA related code.
12881 (rs6000_lra_p): Delete function.
12882 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
12884 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
12886 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
12887 * config/riscv/rtems.h: New file.
12889 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12890 Sudakshina Das <sudi.das@arm.com>
12892 * config/aarch64/aarch64.md
12893 (define_split for and<mode>3nr_compare): Move
12894 non aarch64_logical_operand to a register.
12895 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
12896 register immediate operand to a register.
12897 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
12899 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12901 PR middle-end/81564
12902 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
12904 2017-07-27 Richard Biener <rguenther@suse.de>
12906 PR tree-optimization/81573
12907 PR tree-optimization/81494
12908 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
12909 multi defuse cycle case.
12911 2017-07-27 Richard Biener <rguenther@suse.de>
12913 PR tree-optimization/81571
12914 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
12917 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
12919 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
12920 earlier and only if MASK_FPU is set. Adjust formatting.
12922 2017-07-27 Martin Liska <mliska@suse.cz>
12924 * opt-functions.awk: Add validation of value of Init.
12925 * optc-gen.awk: Pass new argument.
12927 2017-07-27 Martin Liska <mliska@suse.cz>
12929 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
12930 Fix wrong condition.
12932 2017-07-27 Martin Liska <mliska@suse.cz>
12934 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
12935 BBs and edges seen by autoFDO.
12937 2017-07-27 Richard Biener <rguenther@suse.de>
12939 PR tree-optimization/81502
12940 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
12941 with incompatible but same sized type.
12942 (execute_update_addresses_taken): Likewise.
12944 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
12946 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
12947 flag_tree_loop_vectorize rather than flag_tree_vectorize.
12949 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12952 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
12953 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
12954 Change s_operand to memory_operand.
12956 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
12958 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
12959 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
12960 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
12961 Emit instructions rather than returning an expression. Handle TFmode
12962 and KFmode by casting to TImode.
12963 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
12964 (rs6000_emit_le_vsx_store): Likewise.
12965 * config/rs6000/vsx.md (VSX_TI): New iterator.
12966 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
12967 (*vsx_le_undo_permute_<mode>): Likewise.
12968 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
12969 emit the split sequence.
12970 (*vsx_le_perm_store_<mode>): Likewise.
12972 2017-07-27 Jakub Jelinek <jakub@redhat.com>
12974 PR tree-optimization/81555
12975 PR tree-optimization/81556
12976 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
12977 if true, force CHANGED for the recursive invocation.
12978 (reassociate_bb): Remember original length of ops array, pass
12979 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
12981 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
12982 attributes for noipa attribute. For naked attribute use
12983 lookup_attribute first before lookup_attribute_spec.
12984 * final.c (rest_of_handle_final): Disable IPA RA for functions with
12986 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
12987 for functions with noipa attribute.
12988 (cgraph_externally_visible_p): Return true for functions with noipa
12990 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
12991 for functions with noipa attribute.
12992 * doc/extend.texi: Document noipa function attribute.
12993 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
12994 also for functions with noipa attribute.
12995 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
12997 2017-07-26 Andrew Pinski <apinski@cavium.com>
12999 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
13000 vec_unalign_load_cost and vec_unalign_store_cost.
13002 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13004 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
13005 -mvsx-small-integer option.
13006 (ISA_3_0_MASKS_IEEE): Likewise.
13007 (OTHER_VSX_VECTOR_MASKS): Likewise.
13008 (POWERPC_MASKS): Likewise.
13009 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
13010 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
13011 code, only testing for DImode being allowed in non-VSX floating
13013 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
13014 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
13016 (rs6000_option_override_internal): Delete -mvsx-small-integer.
13017 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
13018 TARGET_P8_VECTOR test.
13019 (rs6000_secondary_reload_simple_move): Likewise.
13020 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
13021 since TARGET_P9_VECTOR was already tested.
13022 (rs6000_opt_masks): Remove -mvsx-small-integer.
13023 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
13024 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13026 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
13027 test for TARGET_VEXTRACTUB was used, and that uses
13029 (p9 extract splitter): Likewise.
13030 (vsx_extract_<mode>_di_p9): Likewise.
13031 (vsx_extract_<mode>_store_p9): Likewise.
13032 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
13033 for TARGET_P9_VECTOR was used. Delete code that is now dead with
13034 the elimination of TARGET_VSX_SMALL_INTEGER.
13035 (vsx_extract_<mode>_p8): Likewise.
13036 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
13037 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
13038 (vsx_set_<mode>_p9): Likewise.
13039 (vsx_set_v4sf_p9): Likewise.
13040 (vsx_set_v4sf_p9_zero): Likewise.
13041 (vsx_insert_extract_v4sf_p9): Likewise.
13042 (vsx_insert_extract_v4sf_p9_2): Likewise.
13043 * config/rs6000/rs6000.md (sign extend splitter): Change
13044 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
13045 (floatsi<mode>2_lfiwax_mem): Likewise.
13046 (floatunssi<mode>2_lfiwzx_mem): Likewise.
13047 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
13048 since a test for TARGET_P9_VECTOR was used.
13049 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13050 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
13051 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13052 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13053 TARGET_P8_VECTOR test.
13054 (fix_trunc<mode>si2_stfiwx): Likewise.
13055 (fix_trunc<mode>si2_internal): Likewise.
13056 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
13057 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13059 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13060 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13061 TARGET_P8_VECTOR test.
13062 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13063 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
13064 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13066 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13067 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
13068 since a test for TARGET_P9_VECTOR was used.
13069 (splitter for loading small constants): Likewise.
13071 2017-07-26 Andrew Pinski <apinski@cavium.com>
13073 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
13076 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
13079 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
13080 (fp_valid_at): Likewise.
13082 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13084 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
13085 (qdf24xx_addrcost_table): Likewise.
13086 (cortexa57_tunings): Update to use generic_branch_cost.
13087 (cortexa72_tunings): Likewise.
13088 (cortexa73_tunings): Likewise.
13089 (qdf24xx_tunings): Likewise.
13091 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13093 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
13094 (thunderx2t99_branch_cost): Likewise.
13095 (cortexa35_tunings): Update to use generic_branch_cost.
13096 (cortexa53_tunings): Likewise.
13097 (cortexa57_tunings): Likewise.
13098 (cortexa72_tunings): Likewise.
13099 (cortexa73_tunings): Likewise.
13100 (thunderx2t99_tunings): Likewise.
13102 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13104 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
13105 (sparc_option_override): Honour MASK_FSMULD.
13106 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
13107 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
13108 * config/sparc/sparc.opt (mfsmuld): New option.
13109 * doc/invoke.texi (mfsmuld): Document option.
13111 2017-07-26 Marek Polacek <polacek@redhat.com>
13113 PR middle-end/70992
13114 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
13116 2017-07-26 Richard Biener <rguenther@suse.de>
13118 * gimple-match-head.c (do_valueize): Return OP if valueize
13120 (get_def): New helper to get at the def stmt of a SSA name
13121 if valueize allows.
13122 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
13123 do_valueize to get at the def stmt.
13124 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
13126 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
13128 PR middle-end/46932
13129 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
13131 2017-07-26 Martin Liska <mliska@suse.cz>
13134 * function.c (expand_function_start): Make expansion of
13135 nonlocal_goto_save_area after parm_birth_insn.
13137 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13139 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
13140 from all CPU target flags enable members.
13142 2017-07-26 Richard Biener <rguenther@suse.de>
13144 * genmatch.c (dt_simplify::gen): Make iterator vars const.
13145 (decision_tree::gen): Make 'type' const.
13146 (write_predicate): Likewise.
13148 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
13150 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
13151 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
13152 (rs6000_option_override_internal): Likewise.
13153 (rs6000_expand_vector_set): Likewise.
13154 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
13155 (TARGET_UPPER_REGS_SF): Likewise.
13156 (TARGET_UPPER_REGS_DI): Likewise.
13157 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
13158 (TARGET_DIRECT_MOVE_64BIT): Likewise.
13159 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
13160 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13161 (Splitters for DI constants in Altivec registers): Likewise.
13162 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
13163 (vsx_set_v4sf_p9): Likewise.
13164 (vsx_set_v4sf_p9_zero): Likewise.
13165 (vsx_insert_extract_v4sf_p9): Likewise.
13166 (vsx_insert_extract_v4sf_p9_2): Likewise.
13168 2017-07-25 Carl Love <cel@us.ibm.com>
13170 * doc/extend.texi: Update the built-in documentation file for the
13171 existing built-in functions
13172 vector signed char vec_cnttz (vector signed char);
13173 vector unsigned char vec_cnttz (vector unsigned char);
13174 vector signed short vec_cnttz (vector signed short);
13175 vector unsigned short vec_cnttz (vector unsigned short);
13176 vector signed int vec_cnttz (vector signed int);
13177 vector unsigned int vec_cnttz (vector unsigned int);
13178 vector signed long long vec_cnttz (vector signed long long);
13179 vector unsigned long long vec_cnttz (vector unsigned long long);
13181 2017-07-25 Andrew Pinski <apinski@cavium.com>
13183 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
13184 accesses where the use is for the first operand of a BIT_INSERT.
13186 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
13189 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
13190 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
13192 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
13194 * config/i386/gstabs.h: Delete.
13195 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
13197 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
13199 * config/i386/i386.c (ix86_decompose_address): Do not check for
13200 register RTX when looking at index_reg or base_reg.
13201 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
13203 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
13205 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
13206 to update EH info here.
13208 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13210 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
13212 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13214 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
13216 2017-07-25 Torsten Duwe <duwe@suse.de>
13218 * common.opt: Introduce -fpatchable-function-entry
13219 command line option, and its variables function_entry_patch_area_size
13220 and function_entry_patch_area_start.
13221 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
13222 including a two-value parser.
13223 * target.def (print_patchable_function_entry): New target hook.
13224 * targhooks.h (default_print_patchable_function_entry): New function.
13225 * targhooks.c (default_print_patchable_function_entry): Likewise.
13226 * toplev.c (process_options): Switch off IPA-RA if
13227 patchable function entries are being generated.
13228 * varasm.c (assemble_start_function): Look at the
13229 patchable-function-entry command line switch and current
13230 function attributes and maybe generate NOP instructions by
13231 calling the print_patchable_function_entry hook.
13232 * doc/extend.texi: Document patchable_function_entry attribute.
13233 * doc/invoke.texi: Document -fpatchable_function_entry
13234 command line option.
13235 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
13237 * doc/tm.texi: Re-generate.
13239 2017-07-25 Jakub Jelinek <jakub@redhat.com>
13242 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
13243 TARGET_AVX512DQ rather than TARGET_AVX512BW.
13245 2017-07-25 Tamar Christina <tamar.christina@arm.com>
13247 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
13249 2017-07-25 Richard Biener <rguenther@suse.de>
13251 PR tree-optimization/81455
13252 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
13253 not walk in cycles when looking for guards.
13255 2017-07-25 Richard Biener <rguenther@suse.de>
13257 PR tree-optimization/81529
13258 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
13259 when optimizing backedge uses.
13261 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
13263 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
13265 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
13266 to dl_section_ref. On AIX, append an expression to subtract
13267 the size of the section length to dl_section_ref.
13269 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
13271 * configure.ac: If any of the config.* scripts fail, exit 1.
13272 * configure: Regenerate.
13274 2017-07-25 Richard Biener <rguenther@suse.de>
13276 PR middle-end/81546
13277 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
13278 of immediate uses, be more verbose on errors.
13280 2017-07-25 Richard Biener <rguenther@suse.de>
13282 PR tree-optimization/81510
13283 * tree-vect-loop.c (vect_is_simple_reduction): When the
13284 reduction stmt is not inside the loop bail out.
13286 2017-07-25 Richard Biener <rguenther@suse.de>
13288 PR tree-optimization/81303
13289 * tree-vect-loop-manip.c (vect_loop_versioning): Build
13290 profitability check against LOOP_VINFO_NITERSM1.
13292 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13294 * domwalk.c (cmp_bb_postorder): Simplify.
13295 (sort_bbs_postorder): New function. Use it...
13296 (dom_walker::walk): ...here to optimize common cases.
13298 2017-07-25 Martin Liska <mliska@suse.cz>
13301 * ipa-visibility.c (function_and_variable_visibility): Make the
13302 redirection just on target that supports aliasing.
13303 Fix GNU coding style.
13305 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
13308 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
13310 * config/arm/rtems.h: Likewise.
13311 * config/bfin/rtems.h: Likewise.
13312 * config/i386/rtemself.h: Likewise.
13313 * config/lm32/rtems.h: Likewise.
13314 * config/m32c/rtems.h: Likewise.
13315 * config/m68k/rtemself.h: Likewise.
13316 * config/microblaze/rtems.h: Likewise.
13317 * config/mips/rtems.h: Likewise.
13318 * config/moxie/rtems.h: Likewise.
13319 * config/nios2/rtems.h: Likewise.
13320 * config/powerpcspe/rtems.h: Likewise.
13321 * config/rs6000/rtems.h: Likewise.
13322 * config/rtems.h: Likewise.
13323 * config/sh/rtems.h: Likewise.
13324 * config/sh/rtemself.h: Likewise.
13325 * config/sparc/rtemself.h: Likewise.
13327 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
13330 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
13331 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
13332 * tree-ssa-structalias.c (alias_get_name): Same.
13334 2017-07-25 Bin Cheng <bin.cheng@arm.com>
13337 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
13338 instructions if no du chain is found.
13340 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
13342 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
13344 2017-07-25 Richard Biener <rguenther@suse.de>
13346 PR middle-end/81505
13347 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
13350 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
13352 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
13353 upper-regs options.
13354 (ISA_2_7_MASKS_SERVER): Likewise.
13355 (ISA_3_0_MASKS_IEEE): Likewise.
13356 (OTHER_P8_VECTOR_MASKS): Likewise.
13357 (OTHER_VSX_VECTOR_MASKS): Likewise.
13358 (POWERPC_MASKS): Likewise.
13359 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
13360 duplicate list of options.
13361 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
13362 explicit -mupper-regs options.
13363 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
13364 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
13365 alias for -mupper-regs-df.
13366 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13367 (rs6000_init_hard_regno_mode_ok): Likewise.
13368 (rs6000_option_override_internal): Likewise.
13369 (rs6000_opt_masks): Likewise.
13370 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
13371 options in terms of whether -mvsx or -mpower8-vector was used.
13372 (TARGET_UPPER_REGS_DI): Likewise.
13373 (TARGET_UPPER_REGS_SF): Likewise.
13374 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
13375 -mupper-regs-* options.
13377 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
13379 * passes.c (emergency_dump_function): Print some empty lines and a
13380 header before the RTL dump.
13382 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
13384 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
13386 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
13389 * config/aarch64/aarch64.c (aarch64_classify_symbol):
13390 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
13392 2017-07-24 Carl Love <cel@us.ibm.com>
13394 * config/rs6000/rs6000-c.c: Add support for built-in functions
13395 vector float vec_extract_fp32_from_shorth (vector unsigned short);
13396 vector float vec_extract_fp32_from_shortl (vector unsigned short);
13397 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
13398 vec_extract_fp_from_shortl): Add defines for the two builtins.
13399 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
13400 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
13402 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
13403 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
13404 * doc/extend.texi: Update the built-in documentation file for the
13405 new built-in function.
13407 2017-07-24 Jakub Jelinek <jakub@redhat.com>
13410 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
13412 * doc/generic.texi: Likewise.
13413 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
13414 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
13416 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
13418 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
13419 (aarch64_mls_elt_merge<mode>): Likewise.
13421 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
13423 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13424 having __cxa_atexit.
13426 2017-07-23 Michael Collison <michael.collison@arm.com>
13428 * config/arm/arm.c (arm_option_override): Deprecate
13429 use of -mstructure-size-boundary.
13430 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
13431 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
13433 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13436 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13437 Reduce cost estimate for direct moves.
13439 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
13442 * config/i386/i386.c (ix86_option_override_internal): Disable
13443 BMI, BMI2 and TBM instructions for -m16.
13445 2017-07-21 Carl Love <cel@us.ibm.com>
13447 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13448 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13449 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13450 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13451 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13452 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13453 VMULOSW): New enum "unspec" values.
13454 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13455 altivec_vmulosw): New patterns.
13456 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13457 VMULOSW): Add definitions.
13459 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
13461 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
13462 (qdf24xx): Likewise.
13463 * config/aarch64/aarch64-options-extensions.def (rdma); New.
13464 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
13465 (AARCH64_FL_V8_1): Renumber.
13466 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
13467 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
13468 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
13469 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
13470 rdma to feature modifiers list.
13472 2017-07-21 Yury Gribov <tetra2005@gmail.com>
13474 PR middle-end/56727
13475 * ipa-visibility (function_and_variable_visibility): Convert
13476 recursive PLT call to direct call if appropriate.
13478 2017-07-21 Andrew Pinski <apinski@cavium.com>
13480 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
13481 operand 1 to see if the types precision matches.
13482 * fold-const.c (operand_equal_p): Likewise.
13484 2017-07-21 Richard Biener <rguenther@suse.de>
13486 PR tree-optimization/81303
13487 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
13488 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
13489 (vect_peeling_hash_get_lowest_cost): Adjust.
13490 (vect_enhance_data_refs_alignment): Likewise. Use
13491 vect_get_peeling_costs_all_drs to compute the penalty for no
13492 peeling to match up costs.
13494 2017-07-21 Richard Biener <rguenther@suse.de>
13496 PR tree-optimization/81500
13497 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
13498 we didn't identify a reduction path.
13500 2017-07-21 Tom de Vries <tom@codesourcery.com>
13501 Cesar Philippidis <cesar@codesourcery.com>
13503 PR gcov-profile/81442
13504 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
13507 2017-07-21 Tom de Vries <tom@codesourcery.com>
13510 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
13512 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
13513 nvptx_override_options_after_change.
13515 2017-07-21 Ulrich Drepper <drepper@redhat.com>
13517 * dwarf2out.c (output_file_names): Avoid double testing for
13518 dwarf_version >= 5.
13520 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
13522 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
13524 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13526 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
13528 (try_crossjump_to_edge): Do not punt on partitioned functions.
13530 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13532 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
13533 Put all BBs reachable only via paths crossing cold region to cold
13535 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
13537 2016-07-21 Richard Biener <rguenther@suse.de>
13539 PR tree-optimization/81303
13540 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
13541 into account prologue and epilogue iterations when raising
13542 min_profitable_iters to sth at least covering one vector iteration.
13544 2017-07-21 Tamar Christina <tamar.christina@arm.com>
13546 * config/arm/arm.c (arm_test_cpu_arch_dat):
13549 2017-07-20 Nathan Sidwell <nathan@acm.org>
13551 Remove TYPE_METHODS.
13552 * tree.h (TYPE_METHODS): Delete.
13553 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
13554 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
13555 (dbxout_type_methods): Scan TYPE_FIELDS.
13556 (dbxout_type): Don't check TYPE_METHODS here.
13557 * function.c (use_register_for_decl): Always ignore register for
13558 class types when not optimizing.
13559 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
13560 * tree.c (free_lang_data_in_type): Stitch out member functions and
13561 templates from TYPE_FIELDS.
13562 (build_distinct_type_copy, verify_type_variant,
13563 verify_type): Member fns are on TYPE_FIELDS.
13564 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
13565 * tree-pretty-print.c (dump_generic_node): Likewise.
13567 2017-07-20 Jakub Jelinek <jakub@redhat.com>
13570 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
13571 V2TImode and V4TImode.
13572 (ix86_expand_vector_extract): Likewise.
13573 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
13574 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
13575 (ssescalarmode): Handle V4TImode and V2TImode.
13576 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
13577 (*vec_extractv2ti, *vec_extractv4ti): New insns.
13578 (VEXTRACTI128_MODE): New mode iterator.
13579 (splitter for *vec_extractv?ti first element): New.
13580 (VEC_INIT_MODE): New mode iterator.
13581 (vec_init<mode>): Consolidate 3 expanders into one using
13582 VEC_INIT_MODE mode iterator.
13584 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13586 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
13587 non_spilled_static_chain_regno_p.
13589 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13591 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
13593 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
13595 * bb-reorder.c (connect_traces): Allow copying of blocks within
13598 2017-07-20 Richard Biener <rguenther@suse.de>
13600 * gimple.h (gimple_phi_result): Add gphi * overload.
13601 (gimple_phi_result_ptr): Likewise.
13602 (gimple_phi_arg): Likewise. Adjust index assert to only
13603 allow actual argument accesses rather than all slots available
13605 (gimple_phi_arg_def): Add gphi * overload.
13606 * tree-phinodes.c (make_phi_node): Initialize only actual
13608 (resize_phi_node): Clear memory not covered by old node,
13609 do not initialize excess argument slots.
13610 (reserve_phi_args_for_new_edge): Initialize new argument slot
13613 2017-07-20 Bin Cheng <bin.cheng@arm.com>
13615 PR tree-optimization/81388
13617 2016-07-21 Bin Cheng <bin.cheng@arm.com>
13619 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
13620 by removing computation of may_be_zero.
13622 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13623 Tom de Vries <tom@codesourcery.com>
13625 PR middle-end/81030
13626 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
13627 when gimple level profile disagrees with what RTL expander did.
13629 2017-07-20 Richard Biener <rguenther@suse.de>
13631 PR tree-optimization/61171
13632 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
13633 (vect_analyze_stmt): Add slp instance parameter.
13634 (vectorizable_reduction): Likewise.
13635 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
13636 (vect_is_simple_reduction): Deal with chains not detected
13637 as SLP reduction chain, specifically not properly associated
13638 chains containing a mix of plus/minus.
13639 (get_reduction_op): Remove.
13640 (get_initial_defs_for_reduction): Simplify, pass in whether
13641 this is a reduction chain, pass in the SLP node for the PHIs.
13642 (vect_create_epilog_for_reduction): Get the SLP instance as
13644 (vectorizable_reduction): Get the SLP instance as arg.
13645 During analysis remember the SLP node with the PHIs in the
13646 instance. Simplify getting at the vectorized reduction PHIs.
13647 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
13648 through SLP instance.
13649 (vect_slp_analyze_operations): Likewise.
13650 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
13651 (vect_transform_stmt): Likewise.
13653 2017-07-20 Tom de Vries <tom@codesourcery.com>
13655 PR tree-optimization/81489
13656 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
13657 read of phi arg location to before loop that modifies phi.
13659 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13661 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
13664 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13666 PR middle-end/81331
13667 * except.c (execute): Fix ordering issue.
13669 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13671 PR rtl-optimization/81423
13672 * combine.c (make_compound_operation_int): Don't try to optimize
13673 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
13675 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13677 PR rtl-optimization/81423
13678 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
13679 with a constant that is -1 in the truncated to mode.
13681 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13683 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
13684 (determine_unlikely_bbs): ... here.
13685 * predict.h (propagate_unlikely_bbs_forward): Declare.
13686 * cfgexpand.c (pass_expand::execute): Use it.
13687 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
13689 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
13690 propagate_unlikely_bbs_forward.
13692 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13694 PR middle-end/81331
13695 * except.c (maybe_add_nop_after_section_switch): New function.
13698 2017-07-19 Tom de Vries <tom@codesourcery.com>
13700 * gimple.h (gimple_phi_set_arg): Make assert more strict.
13702 2017-07-19 Tom de Vries <tom@codesourcery.com>
13704 * gimple.h (gimple_phi_arg): Make assert more strict.
13706 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
13708 * config.gcc (powerpc*-*-*): Add mmintrin.h.
13709 * config/rs6000/mmintrin.h: New file.
13710 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
13712 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13714 PR tree-optimization/81346
13715 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
13717 2017-07-19 Tom de Vries <tom@codesourcery.com>
13719 * config/nvptx/nvptx.md (VECIM): Add V2DI.
13721 2017-07-19 Tom de Vries <tom@codesourcery.com>
13723 * config/nvptx/nvptx-modes.def: Add V2DImode.
13724 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
13725 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
13726 (nvptx_output_mov_insn): Handle lack of mov.b128.
13727 (nvptx_print_operand): Handle 'H' and 'L' codes.
13728 (nvptx_vector_mode_supported): Allow V2DImode.
13729 (nvptx_preferred_simd_mode): New function.
13730 (nvptx_data_alignment): New function.
13731 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
13732 nvptx_preferred_simd_mode.
13733 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
13735 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
13737 2017-07-19 Tom de Vries <tom@codesourcery.com>
13739 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
13740 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
13741 (nvptx_vector_mode_supported): New function. Allow V2SImode.
13742 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
13743 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
13744 (mov<VECIM>_insn): New define_insn.
13745 (define_expand "mov<VECIM>): New define_expand.
13747 2017-07-19 Tom de Vries <tom@codesourcery.com>
13749 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
13751 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13753 PR tree-optimization/81346
13754 * fold-const.h (fold_div_compare, range_check_type): Declare.
13755 * fold-const.c (range_check_type): New function.
13756 (build_range_check): Use range_check_type.
13757 (fold_div_compare): No longer static, rewritten into
13758 a match.pd helper function.
13759 (fold_comparison): Don't call fold_div_compare here.
13760 * match.pd (X / C1 op C2): New optimization using fold_div_compare
13761 as helper function.
13763 2017-07-19 Nathan Sidwell <nathan@acm.org>
13765 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
13766 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
13767 * tree.c (find_decls_types_r, verify_type): Use
13768 TYPE_{MIN,MAX}_VALUE_RAW.
13769 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
13770 (hash_tree): Likewise.
13771 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13773 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13776 2017-07-18 Tom de Vries <tom@codesourcery.com>
13778 PR middle-end/81464
13779 * omp-expand.c (expand_omp_for_static_chunk): Handle
13780 equal-argument loop exit phi.
13782 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
13785 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
13786 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
13787 operand 2 predicate.
13788 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
13789 operand 2 predicate.
13790 (ror,rol -> rorx splitters): Use const_int_operand as
13791 operand 2 predicate.
13793 2017-06-18 Richard Biener <rguenther@suse.de>
13795 PR tree-optimization/81410
13796 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
13797 the gap in the ! slp_perm SLP case after each group.
13799 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13801 PR middle-end/81463
13802 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
13805 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13807 PR middle-end/81462
13808 * predict.c (set_even_probabilities): Cleanup; do not affect
13809 probabilities that are already known.
13810 (combine_predictions_for_bb): Call even when count is set.
13812 2017-07-18 Nathan Sidwell <nathan@acm.org>
13814 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
13817 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13820 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
13821 optimization for loop niter analysis.
13823 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
13826 * config/avr/avr.c (avr_optimize_casesi): Don't use
13827 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
13829 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
13831 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
13832 body_cost_vec from _vect_peel_extended_info.
13833 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
13834 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
13837 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13839 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
13841 2017-07-18 Richard Biener <rguenther@suse.de>
13843 PR tree-optimization/80620
13844 PR tree-optimization/81403
13845 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
13846 info when re-using a VN table entry.
13848 2017-07-18 Richard Biener <rguenther@suse.de>
13850 PR tree-optimization/81418
13851 * tree-vect-loop.c (vectorizable_reduction): Properly compute
13852 vectype_in. Verify that with lane-reducing reduction operations
13853 we have a single def-use cycle.
13855 2017-07-17 Carl Love <cel@us.ibm.com>
13857 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
13859 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13860 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13861 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13862 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13863 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13864 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13865 VMULOSW): New enum "unspec" values.
13866 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13867 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13868 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13869 altivec_vmulosw): New patterns.
13870 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13871 VMULOSW): Add definitions.
13872 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
13874 * config/alpha/alpha.c: Include predict.h.
13876 2017-07-17 Yury Gribov <tetra2005@gmail.com>
13878 * tree-vrp.c (compare_assert_loc): Fix comparison function
13879 to return predictable results.
13881 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13883 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
13885 (subdi3): Likewise.
13886 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
13887 * doc/invoke.texi (mexpand-adddi): Update text.
13889 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13891 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
13892 that also clobbers the CC register. The old expand code is moved
13894 (*arc_clzsi2): ... here.
13895 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
13896 the CC register. The old expand code is moved to ...
13897 (arc_ctzsi2): ... here.
13899 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13901 * config/arc/arc.opt (mindexed-loads): Use initial value
13902 TARGET_INDEXED_LOADS_DEFAULT.
13903 (mauto-modify-reg): Use initial value
13904 TARGET_AUTO_MODIFY_REG_DEFAULT.
13905 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
13906 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
13907 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
13908 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
13910 2017-07-17 Martin Liska <mliska@suse.cz>
13913 * opts.c (finish_options): Do not allow -fgnu-tm
13914 w/ -fsanitize={kernel-,}address. Say sorry.
13916 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13919 * tree-loop-distribution.c (classify_partition): Only assert on
13920 numer of iterations.
13921 (merge_dep_scc_partitions): Delete prameter. Update function call.
13922 (distribute_loop): Remove code handling loop with unknown niters.
13923 (pass_loop_distribution::execute): Skip loop with unknown niters.
13925 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13928 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
13929 function sort_partitions_by_post_order.
13931 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13933 PR tree-optimization/81374
13934 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
13935 the max index of basic blocks, rather than number of basic blocks.
13937 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13939 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
13941 (arc_legitimate_pic_operand_p): Likewise.
13942 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
13944 (arc_needs_pcl_p): Likewise.
13945 (arc_legitimate_pc_offset_p): Likewise.
13946 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
13947 function is also used in constrains.md.
13948 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
13949 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
13950 PLUS. Only return true/false in known cases, otherwise assert.
13951 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
13952 is already called in arc_legitimate_constant_p.
13953 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
13955 (LEGITIMATE_PIC_OPERAND_P): Use
13956 arc_raw_symbolic_reference_mentioned_p function.
13957 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
13962 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13963 Andrew Burgess <andrew.burgess@embecosm.com>
13965 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
13966 (arc_return_address_register): New function.
13967 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
13968 (arc_handle_fndecl_attribute): Add naked attribute.
13969 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
13970 (TARGET_WARN_FUNC_RETURN): Likewise.
13971 (arc_allocate_stack_slots_for_args): New function.
13972 (arc_warn_func_return): Likewise.
13973 (machine_function): Change type fn_type.
13974 (arc_compute_function_type): Consider new naked function type,
13975 change function return type.
13976 (arc_must_save_register): Adapt to handle new
13977 arc_compute_function_type's return type.
13978 (arc_expand_prologue): Likewise.
13979 (arc_expand_epilogue): Likewise.
13980 (arc_return_address_regs): Delete.
13981 (arc_return_address_register): New function.
13982 (arc_epilogue_uses): Use above function.
13983 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
13984 (arc_function_type): Change encoding, add naked type.
13985 (ARC_INTERRUPT_P): Change to handle the new encoding.
13986 (ARC_FAST_INTERRUPT_P): Likewise.
13987 (ARC_NORMAL_P): Define.
13988 (ARC_NAKED_P): Likewise.
13989 (arc_compute_function_type): Delete prototype.
13990 * config/arc/arc.md (in_ret_delay_slot): Use
13991 arc_return_address_register function.
13992 (simple_return): Likewise.
13993 (p_return_i): Likewise.
13995 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13997 PR tree-optimization/81428
13998 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
13999 can't be built for those types.
14001 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14003 Remove stuff dead since r239246.
14005 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
14006 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
14007 (avr_inform_devices): Remove dead stuff.
14009 2017-07-17 Tamar Christina <tamar.christina@arm.com>
14011 * config/arm/arm_neon.h: Fix softp typo.
14013 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14015 PR tree-optimization/81365
14016 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
14017 aggregate moves onto bb predecessor edges, make sure there are no
14018 loads that could alias the lhs in between the start of bb and the
14021 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14024 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
14025 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
14026 [LSHIFTRT, outer_code = TRUNCATE]: Same.
14028 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14030 PR tree-optimization/81396
14031 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
14032 (init_symbolic_number): Initialize it to 1.
14033 (perform_symbolic_merge): Add n_ops from both operands into the new
14035 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
14036 without base_addr as useless if they need more than one operation.
14037 (bswap_replace): Handle !bswap case for NULL base_addr.
14039 2017-07-17 Tom de Vries <tom@codesourcery.com>
14042 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
14045 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
14047 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
14048 conditional builtin define __FIX_LEON3FT_B2BST.
14050 2017-07-17 Daniel Cederman <cederman@gaisler.com>
14052 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
14053 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
14056 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14058 PR rtl-optimization/81424
14059 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
14060 to remove potential trapping from operands if -fnon-call-exceptions.
14062 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14064 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
14065 profile_proability for scalling.
14066 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
14068 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14070 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
14072 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14074 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
14075 fixpoint arithmetics.
14077 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14079 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
14080 fixpoint arithmetics.
14082 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14084 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
14085 fixpoint arithmetics.
14087 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14089 * profile-count.h (profile_probability::from_reg_br_prob_note,
14090 profile_probability::to_reg_br_prob_note): New functions.
14091 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
14092 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
14093 * predict.c (probability_reliable_p): Update.
14094 (edge_probability_reliable_p): Update.
14095 (br_prob_note_reliable_p): Update.
14096 (invert_br_probabilities): Update.
14097 (add_reg_br_prob_note): New function.
14098 (combine_predictions_for_insn): Update.
14099 * asan.c (asan_clear_shadow): Update.
14100 * cfgbuild.c (compute_outgoing_frequencies): Update.
14101 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14102 (update_br_prob_note): Update.
14103 (rtl_verify_edges): Update.
14104 (purge_dead_edges): Update.
14105 (fixup_reorder_chain): Update.
14106 * emit-rtl.c (try_split): Update.
14107 * ifcvt.c (cond_exec_process_insns): Update.
14108 (cond_exec_process_if_block): Update.
14109 (dead_or_predicable): Update.
14110 * internal-fn.c (expand_addsub_overflow): Update.
14111 (expand_neg_overflow): Update.
14112 (expand_mul_overflow): Update.
14113 * loop-doloop.c (doloop_modify): Update.
14114 * loop-unroll.c (compare_and_jump_seq): Update.
14115 * optabs.c (emit_cmp_and_jump_insn_1): Update.
14116 * predict.h: Update.
14117 * reorg.c (mostly_true_jump): Update.
14119 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
14120 * config/alpha/alpha.c (emit_unlikely_jump): Update.
14121 * config/arc/arc.c: (emit_unlikely_jump): Update.
14122 * config/arm/arm.c: (emit_unlikely_jump): Update.
14123 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
14124 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
14125 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
14126 (ix86_print_operand): Update.
14127 (ix86_split_fp_branch): Update.
14128 (predict_jump): Update.
14129 * config/ia64/ia64.c (ia64_print_operand): Update.
14130 * config/mmix/mmix.c (mmix_print_operand): Update.
14131 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
14132 (rs6000_expand_split_stack_prologue): Update.
14133 * config/rs6000/rs6000.c: Update.
14134 * config/s390/s390.c (s390_expand_vec_strlen): Update.
14135 (s390_expand_vec_movstr): Update.
14136 (s390_expand_cs_tdsi): Update.
14137 (s390_expand_split_stack_prologue): Update.
14138 * config/sh/sh.c (sh_print_operand): Update.
14139 (expand_cbranchsi4): Update.
14140 (expand_cbranchdi4): Update.
14141 * config/sparc/sparc.c (output_v9branch): Update.
14142 * config/spu/spu.c (get_branch_target): Update.
14143 (ea_load_store_inline): Update.
14144 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
14145 * config/tilepro/tilepro.c: Update.
14147 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14149 * gimplify.c (mostly_copy_tree_r): Revert latest change.
14150 (gimplify_save_expr): Likewise.
14152 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14154 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
14156 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14158 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
14160 * timevar.def (TV_IPA_FNSUMMARY): Define.
14162 2017-07-16 Daniel Cederman <cederman@gaisler.com>
14164 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
14165 to back store errata sensitive sequence from being generated.
14166 (sqrtdf2_fix): Likewise.
14168 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14170 * tree-ssa-threadupdate.c (compute_path_counts,
14171 update_joiner_offpath_counts): Use profile_probability.
14173 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
14176 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14178 * config/arm/arm-c.c (arm_cpu_builtins): Define
14179 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
14181 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
14183 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14184 array entries to represent __ieee128 versions of the
14185 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
14186 scalar_extract_sig, and scalar_insert_exp built-in functions.
14187 (altivec_resolve_overloaded_builtin): Add special case handling
14188 for the __builtin_scalar_insert_exp function, as represented by
14189 the P9V_BUILTIN_VEC_VSIEDP constant.
14190 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
14191 exponent support for __ieee128 argument.
14192 (VSESQP): Add scalar extract signature support for __ieee128
14194 (VSTDCNQP): Add scalar test negative support for __ieee128
14196 (VSIEQP): Add scalar insert exponent support for __int128 argument
14197 with __ieee128 result.
14198 (VSIEQPF): Add scalar insert exponent support for __ieee128
14199 argument with __ieee128 result.
14200 (VSTDCQP): Add scalar test data class support for __ieee128
14202 (VSTDCNQP): Add overload support for scalar test negative with
14203 __ieee128 argument.
14204 (VSTDCQP): Add overload support for scalar test data class
14205 __ieee128 argument.
14206 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
14207 UNSPEC_VSX_SXSIGDP.
14208 (UNSPEC_VSX_SIEXPQP): New constant.
14209 (xsxexpqp): New insn for VSX scalar extract exponent quad
14211 (xsxsigqp): New insn for VSX scalar extract significand quad
14213 (xsiexpqpf): New insn for VSX scalar insert exponent quad
14214 precision with floating point argument.
14215 (xststdcqp): New expand for VSX scalar test data class quad
14217 (xststdcnegqp): New expand for VSX scalar test negative quad
14219 (xststdcqp): New insn to match expansions for VSX scalar test data
14220 class quad precision and VSX scalar test negative quad precision.
14221 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
14222 special case operand checking to enforce that second operand of
14223 VSX scalar test data class with quad precision argument is a 7-bit
14225 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
14226 prototypes and descriptions of __ieee128 versions of
14227 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
14228 scalar_test_data_class, and scalar_test_neg built-in functions.
14230 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14232 PR tree-optimization/81162
14233 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
14234 replace a negate with an add.
14236 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
14238 * doc/invoke.texi (arm/-mcpu): Document +crypto.
14240 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14242 * config/arm/arm-c.c (arm_cpu_builtins): Define
14243 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
14245 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14247 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
14248 (armv8-r): Set ARM Cortex-R52 as default CPU.
14249 * config/arm/arm-tables.opt: Regenerate.
14250 * config/arm/arm-tune.md: Regenerate.
14251 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14253 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
14254 extension for -mcpu=cortex-r52.
14256 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14258 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
14259 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
14260 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
14261 (fp-armv8): Define it as FP_ARMv8 only.
14262 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
14263 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
14265 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
14266 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
14267 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
14268 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
14269 than TARGET_FPU_ARMV8.
14270 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
14271 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
14272 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
14274 * config/arm/neon.md (neon_vrint): Likewise.
14275 (neon_vcvt): Likewise.
14276 (neon_<fmaxmin_op><mode>): Likewise.
14277 (<fmaxmin><mode>3): Likewise.
14278 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
14279 * config/arm/predicates.md (arm_cond_move_operator): Check against
14280 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
14282 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
14284 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
14285 to top of function.
14287 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14289 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
14290 loop in comment with memset.
14292 2017-07-14 Martin Liska <mliska@suse.cz>
14294 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
14295 * dwarf2out.c (is_java): Remove the function.
14296 (output_pubname): Remove usage of the function.
14297 (lower_bound_default): Remove usage of DW_LANG_Java.
14298 (gen_compile_unit_die): Likewise.
14299 * gcc.c: Remove compiler defaults for .java and .zip files.
14300 * gimple-expr.c (remove_suffix): Change as there's no longer
14301 extension than 4-letter one.
14302 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
14303 (gimplify_save_expr): Likewise.
14304 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
14305 as it's possible even for other languages than Java.
14306 * langhooks.h (struct lang_hooks): Remove Java from a comment.
14307 * lto-opts.c (lto_write_options): Remove reference to Java.
14308 * opts.c (strip_off_ending): Update file extension handling.
14309 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
14310 * tree-eh.c (lower_resx): Likewise.
14311 * tree.c (free_lang_data_in_type): Remove dead code.
14312 (find_decls_types_r): Likewise.
14313 (build_common_builtin_nodes): Remove Java from a comment.
14314 (verify_type): Remove dead code.
14315 * varasm.c (assemble_external): Remove Java from a comment.
14317 2017-07-14 Martin Liska <mliska@suse.cz>
14319 * opts.c (finish_options): Add quotes.
14320 (common_handle_option): Likewise.
14322 2017-07-14 Martin Liska <mliska@suse.cz>
14324 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
14325 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
14326 Remove N_SO_PASCAL.
14327 * dwarf2out.c (lower_bound_default): Do not handle
14329 (gen_compile_unit_die): Likewise.
14330 * gcc.c: Remove default extension binding for GNU Pascal.
14331 * stmt.c: Remove Pascal language from a comment.
14332 * xcoffout.c: Likewise.
14334 2017-07-13 David Malcolm <dmalcolm@redhat.com>
14337 * diagnostic-show-locus.c (fixit_cmp): New function.
14338 (layout::layout): Sort m_fixit_hints.
14339 (column_range::column_range): Assert that the values are valid.
14340 (struct char_span): New struct.
14341 (correction::overwrite): New method.
14342 (struct source_line): New struct.
14343 (line_corrections::add_hint): Add assertions. Reimplement memcpy
14344 calls in terms of classes source_line and char_span, and
14345 correction::overwrite.
14346 (selftest::test_overlapped_fixit_printing_2): New function.
14347 (selftest::diagnostic_show_locus_c_tests): Call it.
14349 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
14351 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
14352 early if there is no lhs.
14354 2017-07-13 Martin Liska <mliska@suse.cz>
14356 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
14357 (gen_reference_type_die): Likewise.
14358 * stor-layout.c: Remove Pascal-related comment.
14360 2017-07-13 Martin Liska <mliska@suse.cz>
14362 * opts.c (finish_options): Add quotes to error messages.
14363 (parse_sanitizer_options): Likewise.
14365 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
14367 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
14369 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
14371 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
14373 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
14375 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
14377 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
14379 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
14382 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
14383 provides the hardware capability bits, define the macro
14384 __BUILTIN_CPU_SUPPORTS__.
14385 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
14386 if GLIBC does not provide the hardware capability bits. Add a
14387 gcc_unreachable call if the built-in cpu function is neither
14388 __builtin_cpu_is nor __builtin_cpu_supports.
14389 (rs6000_get_function_versions_dispatcher): Change the warning
14390 that an old GLIBC is used which does not export the capability
14391 bits to be an error.
14392 * doc/extend.texi (target_clones attribute): Document the
14393 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
14394 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
14395 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
14396 the macros defined by GCC if the newer GLIBC is available.
14398 2017-07-12 Jeff Law <law@redhat.com>
14400 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
14401 remaining includes slightly.
14402 * config/riscv/riscv-builtins.c: Include profile-count.h.
14404 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14407 * config/avr/avr.c (avr_set_current_function): In diagnostic
14408 messages: Quote keywords and (parts of) identifiers.
14409 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
14412 2017-07-12 Carl Love <cel@us.ibm.com>
14414 * config/rs6000/rs6000-c.c: Add support for built-in functions
14415 vector bool char vec_revb (vector bool char);
14416 vector bool short vec_revb (vector short char);
14417 vector bool int vec_revb (vector bool int);
14418 vector bool long long vec_revb (vector bool long long);
14419 * doc/extend.texi: Update the built-in documentation file for the
14420 new built-in functions.
14422 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14424 * config/s390/s390.md: Remove movcc splitter.
14426 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14428 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
14429 load/store on condition.
14431 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14434 * config/avr/avr.c (avr_encode_section_info)
14435 [progmem && !TREE_READONLY]: Error if progmem object needs
14438 2017-07-11 Michael Collison <michael.collison@arm.com>
14440 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
14443 2017-07-11 Carl Love <cel@us.ibm.com>
14445 * config/rs6000/rs6000-c.c: Add support for builtins
14446 vector unsigned int vec_parity_lsbb (vector signed int);
14447 vector unsigned int vec_parity_lsbb (vector unsigned int);
14448 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
14449 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
14450 vector unsigned long long vec_parity_lsbb (vector signed long long);
14451 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
14452 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
14453 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
14454 * doc/extend.texi: Update the built-in documentation file for the
14455 new built-in functions.
14457 2017-07-11 David Malcolm <dmalcolm@redhat.com>
14459 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
14460 (layout::m_primary_loc): New field.
14461 (layout::layout): Initialize new field. Move location filtering
14462 logic from here to...
14463 (layout::maybe_add_location_range): ...this new method. Add
14464 support for filtering to just the lines already specified by other
14466 (layout::will_show_line_p): New method.
14467 (gcc_rich_location::add_location_if_nearby): New method.
14468 (selftest::test_add_location_if_nearby): New test function.
14469 (selftest::diagnostic_show_locus_c_tests): Call it.
14470 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
14473 2017-07-11 Tom de Vries <tom@codesourcery.com>
14475 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
14476 (bb_first_real_insn): New function.
14477 (nvptx_single): Add extra initialization of broadcasted condition
14480 2017-07-11 Nathan Sidwell <nathan@acm.org>
14482 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
14484 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
14486 * doc/extend.texi (AVR Function Attributes): Remove weblink to
14487 Binutils doc as TEXI will mess them up.
14488 * doc/invoke.texi (AVR Options): Same here.
14490 2017-07-11 Daniel Cederman <cederman@gaisler.com>
14492 * config/sparc/sparc.opt (mfix-ut700): New option.
14493 (mfix-gr712rc): Likewise.
14494 (sparc_fix_b2bst): New variable.
14495 * doc/invoke.texi (SPARC options): Document them.
14496 (ARM options): Fix warnings.
14497 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
14498 instructions to prevent sequences that can trigger the store-store
14499 errata for certain LEON3FT processors.
14500 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
14501 (sparc_option_override): Set sparc_fix_b2bst appropriately.
14502 * config/sparc/sparc.md (fix_b2bst): New attribute.
14503 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
14505 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
14508 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
14510 (*rsqrtsf2_sse): Ditto.
14512 (div<mode>3): Macroize insn from divdf3 and divsf3
14513 using MODEF mode iterator.
14515 2017-07-10 Martin Sebor <msebor@redhat.com>
14517 PR tree-optimization/80397
14518 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
14519 instead of testing for equality to INTEGER_TYPE.
14521 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
14523 * config.gcc: Remove uclibc from arc target spec.
14525 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
14527 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
14529 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14532 * lto-wrapper.c (remove_option): New function.
14533 (merge_and_complain): Merge PIC/PIE options more realistically.
14535 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14537 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
14541 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
14542 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
14543 * config.in: Regenerate.
14544 * configure: Regenerate.
14545 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
14546 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
14547 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
14548 (TARGET_GASISR_PROLOGUES): ...target mask.
14549 * common/config/avr/avr-common.c
14550 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
14551 Set -mgas-isr-prologues.
14552 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
14553 INSERT_PASS_BEFORE for it.
14554 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
14555 * config/avr/avr.c (avr_option_override)
14556 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
14557 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
14558 (avr_attribute_table) <no_gccisr>: Add new function attribute.
14559 (avr_set_current_function) <is_no_gccisr>: Init machine field.
14560 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
14562 (make_avr_pass_pre_proep): New function.
14563 (emit_push_sfr) <treg>: Add argument to function and use it
14564 instead of TMP_REG.
14565 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
14566 and set machine->gasisr.yes.
14567 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
14568 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
14569 __gcc_isr.n_pushed to .L__stack_usage.
14570 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
14571 (avr_asm_final_postscan_insn): ...this new static function.
14572 * config/avr/avr.h (machine_function)
14573 <is_no_gccisr, use_L__stack_usage>: New fields.
14574 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
14575 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
14576 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
14577 (gasisr, *gasisr): New expander and insn.
14578 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14579 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
14580 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
14582 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
14584 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
14587 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14589 Move jump-tables out of .text again.
14592 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
14593 (ASM_OUTPUT_ADDR_VEC): New function.
14594 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
14595 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
14596 INSN_ADDRESSes as asm comment.
14597 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
14598 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
14599 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
14600 * config/avr/avr.md (*tablejump): Adjust comment.
14601 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
14602 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
14604 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
14605 (avr_output_addr_vec): New proto.
14606 (avr_log_t) <insn_addresses>: New field.
14608 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
14611 * config/i386/i386.c (ix86_function_arg_advance): Set
14612 outgoing_args_on_stack to true if there are outgoing arguments
14614 (ix86_function_arg): Likewise.
14615 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
14616 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
14617 * config/i386/i386.h (machine_function): Add
14618 outgoing_args_on_stack.
14620 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
14622 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14623 supporting pthreds.
14624 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
14626 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
14628 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
14629 (REAL_H): Remove $(MACHMODE_H).
14630 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
14632 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
14633 $(MACHMODE_H) and double-int.h.
14634 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
14636 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
14639 2017-07-07 Andrew Pinski <apinski@cavium.com>
14641 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
14642 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
14644 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14646 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
14647 Add warning if GCC was not configured to link against a GLIBC that
14648 exports the hardware capability bits.
14649 (make_resolver_func): Make resolver function private and not a
14650 COMDAT function. Create the name with clone_function_name instead
14651 of make_unique_name.
14654 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
14655 correct operand in doing the split.
14657 2017-07-07 Carl Love <cel@us.ibm.com>
14659 * config/rs6000/rs6000-c: Add support for built-in function
14660 vector unsigned short vec_pack_to_short_fp32 (vector float,
14662 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
14663 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
14664 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
14665 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
14666 (convert_4f32_8i16): Add define_expand.
14667 * doc/extend.texi: Update the built-in documentation file for the
14668 new built-in function.
14670 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14672 * config/sparc/m8.md: New file.
14673 * config/sparc/sparc.md: Include m8.md.
14675 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14677 * config/sparc/sparc.opt: New option -mvis4b.
14678 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
14679 (sparc_option_override): Handle VIS4B.
14680 (enum sparc_builtins): Define
14681 SPARC_BUILTIN_DICTUNPACK{8,16,32},
14682 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
14683 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
14684 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
14685 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
14686 (check_constant_argument): New function.
14687 (sparc_vis_init_builtins): Define builtins
14688 __builtin_vis_dictunpack{8,16,32},
14689 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
14690 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
14691 __builtin_vis_fpcmpde{8,16,32}shl and
14692 __builtin_vis_fpcmpur{8,16,32}shl.
14693 (sparc_expand_builtin): Check that the constant operands to
14694 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
14695 constant and in range.
14696 * config/sparc/sparc-c.c (sparc_target_macros): Handle
14698 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
14699 (SPARC_IMM5_P): Likewise.
14700 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
14701 (enabled): Handle vis4b.
14702 (UNSPEC_DICTUNPACK): New unspec.
14703 (UNSPEC_FPCMPSHL): Likewise.
14704 (UNSPEC_FPUCMPSHL): Likewise.
14705 (UNSPEC_FPCMPDESHL): Likewise.
14706 (UNSPEC_FPCMPURSHL): Likewise.
14707 (cpu_feature): New CPU feature `vis4b'.
14708 (dictunpack{8,16,32}): New insns.
14709 (FPCSMODE): New mode iterator.
14710 (fpcscond): New code iterator.
14711 (fpcsucond): Likewise.
14712 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
14713 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
14714 (fpcmpde{8,16,32}{si,di}shl): Likewise.
14715 (fpcmpur{8,16,32}{si,di}shl): Likewise.
14716 * config/sparc/constraints.md: Define constraints `q' for unsigned
14717 2-bit integer constants and `t' for unsigned 5-bit integer
14719 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
14721 (imm5_operand_dictunpack16): Likewise.
14722 (imm5_operand_dictunpack32): Likewise.
14723 (imm2_operand): Likewise.
14724 * doc/invoke.texi (SPARC Options): Document -mvis4b.
14725 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
14726 ditunpack* and fpcmp*shl builtins.
14728 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14730 * config.gcc: Handle m8 in --with-{cpu,tune} options.
14731 * config.in: Add HAVE_AS_SPARC6 define.
14732 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
14734 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
14736 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
14737 (CPP_CPU_SPEC): Handle m8.
14738 (ASM_CPU_SPEC): Likewise.
14739 * config/sparc/sparc-opts.h (enum processor_type): Add
14741 * config/sparc/sparc.c (m8_costs): New struct.
14742 (sparc_option_override): Handle TARGET_CPU_m8.
14743 (sparc32_initialize_trampoline): Likewise.
14744 (sparc64_initialize_trampoline): Likewise.
14745 (sparc_issue_rate): Likewise.
14746 (sparc_register_move_cost): Likewise.
14747 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
14748 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
14749 (ASM_CPU64_DEFAULT_SPEC): Likewise.
14750 (CPP_CPU_SPEC): Handle M8.
14751 (ASM_CPU_SPEC): Likewise.
14752 (AS_M8_FLAG): Define.
14753 * config/sparc/sparc.md: Add m8 to the cpu attribute.
14754 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
14755 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
14757 * configure: Regenerate.
14758 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
14761 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14763 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
14765 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
14766 ("*movdi_insn_sp32"): Do not set v3pipe.
14767 ("*movsi_insn"): Likewise.
14768 ("*movdi_insn_sp64"): Likewise.
14769 ("*movsf_insn"): Likewise.
14770 ("*movdf_insn_sp32"): Likewise.
14771 ("*movdf_insn_sp64"): Likewise.
14772 ("*zero_extendsidi2_insn_sp64"): Likewise.
14773 ("*sign_extendsidi2_insn"): Likewise.
14774 ("*mov<VM32:mode>_insn"): Likewise.
14775 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14776 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14777 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14778 ("<vlop:code><VL:mode>3"): Likewise.
14779 ("*not_<vlop:code><VL:mode>3"): Likewise.
14780 ("*nand<VL:mode>_vis"): Likewise.
14781 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
14782 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
14783 ("one_cmpl<VL:mode>2"): Likewise.
14784 ("faligndata<VM64:mode>_vis"): Likewise.
14785 ("alignaddrsi_vis"): Likewise.
14786 ("alignaddrdi_vis"): Likweise.
14787 ("alignaddrlsi_vis"): Likewise.
14788 ("alignaddrldi_vis"): Likewise.
14789 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14790 ("bmaskdi_vis"): Likewise.
14791 ("bmasksi_vis"): Likewise.
14792 ("bshuffle<VM64:mode>_vis"): Likewise.
14793 ("cmask8<P:mode>_vis"): Likewise.
14794 ("cmask16<P:mode>_vis"): Likewise.
14795 ("cmask32<P:mode>_vis"): Likewise.
14796 ("pdistn<P:mode>_vis"): Likewise.
14797 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14799 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14801 * config/sparc/sparc.md ("subtype"): New insn attribute.
14802 ("*wrgsr_sp64"): Set insn subtype.
14803 ("*rdgsr_sp64"): Likewise.
14804 ("alignaddrsi_vis"): Likewise.
14805 ("alignaddrdi_vis"): Likewise.
14806 ("alignaddrlsi_vis"): Likewise.
14807 ("alignaddrldi_vis"): Likewise.
14808 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14809 ("fexpand_vis"): Likewise.
14810 ("fpmerge_vis"): Likewise.
14811 ("faligndata<VM64:mode>_vis"): Likewise.
14812 ("bshuffle<VM64:mode>_vis"): Likewise.
14813 ("cmask8<P:mode>_vis"): Likewise.
14814 ("cmask16<P:mode>_vis"): Likewise.
14815 ("cmask32<P:mode>_vis"): Likewise.
14816 ("fchksm16_vis"): Likewise.
14817 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
14818 ("fmean16_vis"): Likewise.
14819 ("fp<plusminus_insn>64_vis"): Likewise.
14820 ("<plusminus_insn>v8qi3"): Likewise.
14821 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14822 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
14823 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
14824 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
14825 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
14826 ("*movqi_insn"): Likewise.
14827 ("*movhi_insn"): Likewise.
14828 ("*movsi_insn"): Likewise.
14829 ("movsi_pic_gotdata_op"): Likewise.
14830 ("*movdi_insn_sp32"): Likewise.
14831 ("*movdi_insn_sp64"): Likewise.
14832 ("movdi_pic_gotdata_op"): Likewise.
14833 ("*movsf_insn"): Likewise.
14834 ("*movdf_insn_sp32"): Likewise.
14835 ("*movdf_insn_sp64"): Likewise.
14836 ("*zero_extendhisi2_insn"): Likewise.
14837 ("*zero_extendqihi2_insn"): Likewise.
14838 ("*zero_extendqisi2_insn"): Likewise.
14839 ("*zero_extendqidi2_insn"): Likewise.
14840 ("*zero_extendhidi2_insn"): Likewise.
14841 ("*zero_extendsidi2_insn_sp64"): Likewise.
14842 ("ldfsr"): Likewise.
14843 ("prefetch_64"): Likewise.
14844 ("prefetch_32"): Likewise.
14845 ("tie_ld32"): Likewise.
14846 ("tie_ld64"): Likewise.
14847 ("*tldo_ldub_sp32"): Likewise.
14848 ("*tldo_ldub1_sp32"): Likewise.
14849 ("*tldo_ldub2_sp32"): Likewise.
14850 ("*tldo_ldub_sp64"): Likewise.
14851 ("*tldo_ldub1_sp64"): Likewise.
14852 ("*tldo_ldub2_sp64"): Likewise.
14853 ("*tldo_ldub3_sp64"): Likewise.
14854 ("*tldo_lduh_sp32"): Likewise.
14855 ("*tldo_lduh1_sp32"): Likewise.
14856 ("*tldo_lduh_sp64"): Likewise.
14857 ("*tldo_lduh1_sp64"): Likewise.
14858 ("*tldo_lduh2_sp64"): Likewise.
14859 ("*tldo_lduw_sp32"): Likewise.
14860 ("*tldo_lduw_sp64"): Likewise.
14861 ("*tldo_lduw1_sp64"): Likewise.
14862 ("*tldo_ldx_sp64"): Likewise.
14863 ("*mov<VM32:mode>_insn"): Likewise.
14864 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14865 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14867 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14869 * config/sparc/sparc.md ("type"): New insn type viscmp.
14870 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
14872 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
14873 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
14874 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14875 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
14877 ("n7_vis_logical_11cycle"): Likewise.
14878 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
14879 * config/sparc/niagara2.md ("niag3_vis": Likewise.
14880 * config/sparc/niagara.md ("niag_vis"): Likewise.
14881 * config/sparc/ultra3.md ("us3_fga"): Likewise.
14882 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
14884 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14886 * config/sparc/sparc.md: New instruction type `bmask'.
14887 (bmaskdi_vis): Use the `bmask' type.
14888 (bmasksi_vis): Likewise.
14889 * config/sparc/ultra3.md (us3_array): Likewise.
14890 * config/sparc/niagara7.md (n7_array): Likewise.
14891 * config/sparc/niagara4.md (n4_array): Likewise.
14892 * config/sparc/niagara2.md (niag2_vis): Likewise.
14893 (niag3_vis): Likewise.
14894 * config/sparc/niagara.md (niag_vis): Likewise.
14896 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14898 * ipa-comdats.c: Remove optimize check from gate.
14899 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
14900 for functions not optimized.
14901 (ipa_fn_summary_read): Skip optimize check.
14902 (ipa_fn_summary_write): Likewise.
14903 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
14905 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
14907 (can_inline_edge_p): Check flag_pcc_struct_return for match.
14908 (check_callers): Give up on caller which is not optimized.
14909 (inline_small_functions): Likewise.
14910 (ipa_inline): Do not give up when not optimizing.
14911 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
14912 away unoptimizes cdtors.
14913 (whole_program_function_and_variable_visibility): Do
14914 ipa_discover_readonly_nonaddressable_vars in LTO mode.
14915 * ipa.c (process_references): Do not check optimize.
14916 (symbol_table::remove_unreachable_nodes): Update optimize check.
14917 (set_writeonly_bit): Update optimize check.
14918 (pass_ipa_cdtor_merge::gate): Do not check optimize.
14919 (pass_ipa_single_use::gate): Remove.
14921 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14923 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
14924 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
14925 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
14926 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
14927 permute_load, permute_store, adjust_extract, adjust_splat,
14928 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
14929 replace_swap_with_copy, dump_swap_insn_table,
14930 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
14931 recombine_lvx_pattern, recombine_stvx_pattern,
14932 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
14933 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
14934 to file rs6000-p8swap.c.
14935 * config/rs6000/rs6000-p8swap.c: New file.
14936 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
14937 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
14938 and rs6000*-*-* targets.
14940 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14942 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
14944 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14946 * lto-wrapper.c (merge_and_complain): Do not merge
14947 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
14948 fsigned_zeros, ftrapping_math, fwrapv.
14949 (append_compiler_options): Do not track these options.
14950 (append_linker_options): Likewie
14952 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14954 * cgraphunit.c (cgraph_node::finalize_function): When
14955 !flag_toplevel_reorde set no_reorder flag.
14956 (varpool_node::finalize_decl): Likewise.
14957 (symbol_table::compile): Drop no toplevel reorder path.
14959 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14961 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
14962 edges; zero probability is not better than uninitialized.
14964 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
14966 * asan.h (asan_sanitize_allocas_p): Declare.
14967 * asan.c (asan_sanitize_allocas_p): New function.
14968 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
14969 (handle_builtin_alloca): Likewise.
14970 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
14971 if !asan_sanitize_allocas_p.
14972 * params.def (asan-instrument-allocas): Add new option.
14973 * params.h (ASAN_PROTECT_ALLOCAS): Define.
14974 * opts.c (common_handle_option): Disable allocas sanitization for
14977 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
14979 * asan.c: Include gimple-fold.h.
14980 (get_last_alloca_addr): New function.
14981 (handle_builtin_stackrestore): Likewise.
14982 (handle_builtin_alloca): Likewise.
14983 (asan_emit_allocas_unpoison): Likewise.
14984 (get_mem_refs_of_builtin_call): Add new parameter, remove const
14985 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
14986 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
14987 (instrument_builtin_call): Pass gimple iterator to
14988 get_mem_refs_of_builtin_call.
14989 (last_alloca_addr): New global.
14990 * asan.h (asan_emit_allocas_unpoison): Declare.
14991 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
14992 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
14993 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
14994 if function calls alloca.
14995 * gimple-fold.c (replace_call_with_value): Remove static keyword.
14996 * gimple-fold.h (replace_call_with_value): Declare.
14997 * internal-fn.c: Include asan.h.
14998 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
14999 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
15001 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15003 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
15004 (C_SELFTEST_FLAGS): New.
15005 (CPP_SELFTEST_FLAGS): New.
15006 (SELFTEST_DEPS): New, from deps of s-selftest.
15007 (C_SELFTEST_DEPS): New, from deps of s-selftest.
15008 (CPP_SELFTEST_DEPS): New.
15009 (selftest): Add dependency on s-selftest-c++.
15010 (s-selftest): Rename to...
15011 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
15012 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
15013 than SELFTEST_FLAGS.
15014 (selftest-gdb): Rename to...
15015 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
15017 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
15018 (selftest-valgrind): Rename to...
15019 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
15021 (selftest-valgrind): Reintroduce as an alias for
15022 selftest-c-valgrind.
15023 (s-selftest-c++): New.
15024 (selftest-c++-gdb): New.
15025 (selftest-c++-valgrind): New.
15027 2017-07-06 Olivier Hainque <hainque@adacore.com>
15029 * gcc.c (process_command): When deciding if undefined variables
15030 should be ignored when processing specs, accept "gcc -v" as well.
15032 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15034 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
15035 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
15037 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15039 * config/arm/arm-cpus.in (armv8-r): Add new entry.
15040 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
15041 * config/arm/arm-tables.opt: Regenerate.
15042 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
15044 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
15046 2017-07-06 Carl Love <cel@us.ibm.com>
15048 * ChangeLog: Clean up from mid air collision
15050 2017-07-06 Carl Love <cel@us.ibm.com>
15052 * config/rs6000/rs6000-c.c: Add support for built-in functions
15053 vector signed int vec_subc (vector signed int, vector signed int);
15054 vector signed __int128 vec_subc (vector signed __int128,
15055 vector signed __int128);
15056 vector unsigned __int128 vec_subc (vector unsigned __int128,
15057 vector unsigned __int128);
15058 vector signed int vec_sube (vector signed int, vector signed int,
15059 vector signed int);
15060 vector unsigned int vec_sube (vector unsigned int,
15061 vector unsigned int,
15062 vector unsigned int);
15063 vector signed __int128 vec_sube (vector signed __int128,
15064 vector signed __int128,
15065 vector signed__int128);
15066 vector unsigned __int128 vec_sube (vector unsigned __int128,
15067 vector unsigned __int128,
15068 vector unsigned __int128);
15069 vector signed int vec_subec (vector signed int, vector signed int,
15070 vector signed int);
15071 vector unsigned int vec_subec (vector unsigned int,
15072 vector unsigned int,
15073 vector unsigned int);
15074 vector signed __int128 vec_subec (vector signed __int128,
15075 vector signed __int128,
15076 vector signed__int128);
15077 vector unsigned __int128 vec_subec (vector unsigned __int128,
15078 vector unsigned __int128,
15079 vector unsigned __int128);
15080 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
15081 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
15082 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
15083 BU_ALTIVEC_OVERLOAD_X definitions.
15084 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
15085 * doc/extend.texi: Update the built-in documentation file for the new
15086 built-in functions.
15088 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15091 * diagnostic-show-locus.c (layout::layout): Use start and finish
15092 spelling location for the start and finish of each range.
15093 * genmatch.c (linemap_client_expand_location_to_spelling_point):
15094 Add unused aspect param.
15095 * input.c (expand_location_1): Add "aspect" param, and use it
15096 to access the correct part of the location.
15097 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
15099 (expand_location_to_spelling_point): Likewise.
15100 (linemap_client_expand_location_to_spelling_point): Add "aspect"
15101 param, and pass it to expand_location_1.
15103 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
15105 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
15106 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
15107 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
15108 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
15109 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
15110 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
15111 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
15112 _mm_maskz_getmant_ss): New intrinsics.
15113 (__builtin_ia32_getexpss128_mask): Changed to ...
15114 __builtin_ia32_getexpss128_round ... this.
15115 (__builtin_ia32_getexpsd128_mask): Changed to ...
15116 __builtin_ia32_getexpsd128_round ... this.
15117 * config/i386/i386-builtin-types.def
15118 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
15119 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
15120 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
15121 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
15122 __builtin_ia32_getmantss_mask_round): New builtins.
15123 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
15124 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
15125 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
15126 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
15127 * config/i386/sse.md
15128 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
15129 avx512f_sgetexp<mode><mask_scalar_name>
15130 <round_saeonly_scalar_name> ... this.
15131 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
15132 %0, %1, %2<round_saeonly_op3>}): Changed to ...
15133 vgetexp<ssescalarmodesuffix>
15134 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15135 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
15136 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
15137 avx512f_vgetmant<mode><mask_scalar_name>
15138 <round_saeonly_scalar_name> ... this.
15139 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
15140 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
15141 vgetmant<ssescalarmodesuffix>
15142 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
15143 %0<mask_scalar_operand4>, %1, %2
15144 <round_saeonly_scalar_mask_op4>, %3} ... this.
15145 * config/i386/subst.md (mask_scalar_operand4,
15146 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
15147 round_saeonly_scalar_nimm_predicate): New subst attributes.
15149 2017-07-06 Julia Koval <julia.koval@intel.com>
15151 * config/i386/i386.c (ix86_erase_embedded_rounding):
15152 Remove code for old rounding pattern.
15154 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
15156 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
15158 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
15160 * doc/sourcebuild.texi (Test Directives, Variants of
15161 dg-require-support): Add documentation for dg-require-stack-check.
15163 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
15165 * config/i386/subst.md (mask_scalar, round_scalar,
15166 round_saeonly_scalar): New meta-templates.
15167 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
15168 round_scalar_mask_operand3, round_scalar_mask_op3,
15169 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
15170 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
15171 round_saeonly_scalar_constraint,
15172 round_saeonly_scalar_prefix): New subst attribute.
15173 * config/i386/sse.md
15174 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
15175 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
15176 <round_scalar_name> ... this.
15177 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
15178 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
15179 <round_scalar_name> ... this.
15180 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
15181 <sse>_vm<code><mode>3<mask_scalar_name>
15182 <round_saeonly_scalar_name> ... this.
15183 (v<plusminus_mnemonic><ssescalarmodesuffix>
15184 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
15185 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
15186 v<plusminus_mnemonic><ssescalarmodesuffix>
15187 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15188 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
15189 (v<multdiv_mnemonic><ssescalarmodesuffix>
15190 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
15191 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
15192 v<multdiv_mnemonic><ssescalarmodesuffix>
15193 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15194 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
15195 (v<maxmin_float><ssescalarmodesuffix>
15196 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
15197 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
15198 v<maxmin_float><ssescalarmodesuffix>
15199 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15200 %0<mask_scalar_operand3>, %1, %<iptr>2
15201 <round_saeonly_scalar_mask_op3>} ... this.
15203 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
15205 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
15206 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
15208 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
15209 Alan Hayward <alan.hayward@arm.com>
15210 David Sherwood <david.sherwood@arm.com>
15212 * combine.c (simplify_if_then_else): Remove "enum" before
15214 * compare-elim.c (can_eliminate_compare): Likewise.
15215 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15217 (aarch64_lookup_simd_builtin_type): Likewise.
15218 (aarch64_simd_builtin_type): Likewise.
15219 (aarch64_init_simd_builtin_types): Likewise.
15220 (aarch64_simd_expand_args): Likewise.
15221 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
15223 (aarch64_reverse_mask): Likewise.
15224 (aarch64_simd_emit_reg_reg_move): Likewise.
15225 (aarch64_gen_adjusted_ldpstp): Likewise.
15226 (aarch64_ccmp_mode_to_code): Likewise.
15227 (aarch64_operands_ok_for_ldpstp): Likewise.
15228 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15229 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15231 (aarch64_min_divisions_for_recip_mul): Likewise.
15232 (aarch64_reassociation_width): Likewise.
15233 (aarch64_get_condition_code_1): Likewise.
15234 (aarch64_simd_emit_reg_reg_move): Likewise.
15235 (aarch64_simd_attr_length_rglist): Likewise.
15236 (aarch64_reverse_mask): Likewise.
15237 (aarch64_operands_ok_for_ldpstp): Likewise.
15238 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15239 (aarch64_gen_adjusted_ldpstp): Likewise.
15240 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
15242 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
15243 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
15244 (arm_lookup_simd_builtin_type): Likewise.
15245 (arm_simd_builtin_type): Likewise.
15246 (arm_init_simd_builtin_types): Likewise.
15247 (arm_expand_builtin_args): Likewise.
15248 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
15249 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
15250 (ft32_setup_incoming_varargs): Likewise.
15251 (ft32_function_arg): Likewise.
15252 (ft32_function_arg_advance): Likewise.
15253 (ft32_pass_by_reference): Likewise.
15254 (ft32_arg_partial_bytes): Likewise.
15255 (ft32_valid_pointer_mode): Likewise.
15256 (ft32_addr_space_pointer_mode): Likewise.
15257 (ft32_addr_space_legitimate_address_p): Likewise.
15258 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
15260 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
15261 (ix86_emit_outlined_ms2sysv_restore): Likewise.
15262 (iamcu_alignment): Likewise.
15263 (canonicalize_vector_int_perm): Likewise.
15264 (ix86_noce_conversion_profitable_p): Likewise.
15265 (ix86_mpx_bound_mode): Likewise.
15266 (ix86_operands_ok_for_move_multiple): Likewise.
15267 * config/microblaze/microblaze-protos.h
15268 (microblaze_expand_conditional_branch_reg): Likewise.
15269 * config/microblaze/microblaze.c
15270 (microblaze_expand_conditional_branch_reg): Likewise.
15271 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15273 (rs6000_reassociation_width): Likewise.
15274 (rs6000_invalid_binary_op): Likewise.
15275 (fusion_p9_p): Likewise.
15276 (emit_fusion_p9_load): Likewise.
15277 (emit_fusion_p9_store): Likewise.
15278 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
15280 (riscv_hard_regno_mode_ok_p): Likewise.
15281 (riscv_address_insns): Likewise.
15282 (riscv_split_symbol): Likewise.
15283 (riscv_legitimize_move): Likewise.
15284 (riscv_function_value): Likewise.
15285 (riscv_hard_regno_nregs): Likewise.
15286 (riscv_expand_builtin): Likewise.
15287 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
15288 (riscv_build_integer): Likewise.
15289 (riscv_split_integer): Likewise.
15290 (riscv_legitimate_constant_p): Likewise.
15291 (riscv_cannot_force_const_mem): Likewise.
15292 (riscv_regno_mode_ok_for_base_p): Likewise.
15293 (riscv_valid_base_register_p): Likewise.
15294 (riscv_valid_offset_p): Likewise.
15295 (riscv_valid_lo_sum_p): Likewise.
15296 (riscv_classify_address): Likewise.
15297 (riscv_legitimate_address_p): Likewise.
15298 (riscv_address_insns): Likewise.
15299 (riscv_load_store_insns): Likewise.
15300 (riscv_force_binary): Likewise.
15301 (riscv_split_symbol): Likewise.
15302 (riscv_force_address): Likewise.
15303 (riscv_legitimize_address): Likewise.
15304 (riscv_move_integer): Likewise.
15305 (riscv_legitimize_const_move): Likewise.
15306 (riscv_legitimize_move): Likewise.
15307 (riscv_address_cost): Likewise.
15308 (riscv_subword): Likewise.
15309 (riscv_output_move): Likewise.
15310 (riscv_canonicalize_int_order_test): Likewise.
15311 (riscv_emit_int_order_test): Likewise.
15312 (riscv_function_arg_boundary): Likewise.
15313 (riscv_pass_mode_in_fpr_p): Likewise.
15314 (riscv_pass_fpr_single): Likewise.
15315 (riscv_pass_fpr_pair): Likewise.
15316 (riscv_get_arg_info): Likewise.
15317 (riscv_function_arg): Likewise.
15318 (riscv_function_arg_advance): Likewise.
15319 (riscv_arg_partial_bytes): Likewise.
15320 (riscv_function_value): Likewise.
15321 (riscv_pass_by_reference): Likewise.
15322 (riscv_setup_incoming_varargs): Likewise.
15323 (riscv_print_operand): Likewise.
15324 (riscv_elf_select_rtx_section): Likewise.
15325 (riscv_save_restore_reg): Likewise.
15326 (riscv_for_each_saved_reg): Likewise.
15327 (riscv_register_move_cost): Likewise.
15328 (riscv_hard_regno_mode_ok_p): Likewise.
15329 (riscv_hard_regno_nregs): Likewise.
15330 (riscv_class_max_nregs): Likewise.
15331 (riscv_memory_move_cost): Likewise.
15332 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
15333 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
15334 (rl78_addr_space_address_mode): Likewise.
15335 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15337 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15338 (rs6000_reassociation_width): Likewise.
15339 (rs6000_invalid_binary_op): Likewise.
15340 (fusion_p9_p): Likewise.
15341 (emit_fusion_p9_load): Likewise.
15342 (emit_fusion_p9_store): Likewise.
15343 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
15344 (ok_for_simple_move_operands): Likewise.
15345 (ok_for_simple_move_strict_operands): Likewise.
15346 (ok_for_simple_arith_logic_operands): Likewise.
15347 (visium_legitimize_reload_address): Likewise.
15348 (visium_select_cc_mode): Likewise.
15349 (output_cbranch): Likewise.
15350 (visium_split_double_move): Likewise.
15351 (visium_expand_copysign): Likewise.
15352 (visium_expand_int_cstore): Likewise.
15353 (visium_expand_fp_cstore): Likewise.
15354 * config/visium/visium.c (visium_pass_by_reference): Likewise.
15355 (visium_function_arg): Likewise.
15356 (visium_function_arg_advance): Likewise.
15357 (visium_libcall_value): Likewise.
15358 (visium_setup_incoming_varargs): Likewise.
15359 (visium_legitimate_constant_p): Likewise.
15360 (visium_legitimate_address_p): Likewise.
15361 (visium_legitimize_address): Likewise.
15362 (visium_secondary_reload): Likewise.
15363 (visium_register_move_cost): Likewise.
15364 (visium_memory_move_cost): Likewise.
15365 (prepare_move_operands): Likewise.
15366 (ok_for_simple_move_operands): Likewise.
15367 (ok_for_simple_move_strict_operands): Likewise.
15368 (ok_for_simple_arith_logic_operands): Likewise.
15369 (visium_function_value_1): Likewise.
15370 (rtx_ok_for_offset_p): Likewise.
15371 (visium_legitimize_reload_address): Likewise.
15372 (visium_split_double_move): Likewise.
15373 (visium_expand_copysign): Likewise.
15374 (visium_expand_int_cstore): Likewise.
15375 (visium_expand_fp_cstore): Likewise.
15376 (visium_split_cstore): Likewise.
15377 (visium_select_cc_mode): Likewise.
15378 (visium_split_cbranch): Likewise.
15379 (output_cbranch): Likewise.
15380 (visium_print_operand_address): Likewise.
15381 * expmed.c (flip_storage_order): Likewise.
15382 * expmed.h (emit_cstore): Likewise.
15383 (flip_storage_order): Likewise.
15384 * genrecog.c (validate_pattern): Likewise.
15385 * hsa-gen.c (gen_hsa_addr): Likewise.
15386 * internal-fn.c (expand_arith_overflow): Likewise.
15387 * ira-color.c (allocno_copy_cost_saving): Likewise.
15388 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15389 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
15390 (process_invariant_for_inheritance): Likewise.
15391 * lra-eliminations.c (move_plus_up): Likewise.
15392 * omp-low.c (lower_oacc_reductions): Likewise.
15393 * simplify-rtx.c (simplify_subreg): Likewise.
15394 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
15395 (TARGET_CHKP_BOUND_MODE): Likewise..
15396 * targhooks.c (default_chkp_bound_mode): Likewise.
15397 (default_setup_incoming_vararg_bounds): Likewise.
15398 * targhooks.h (default_chkp_bound_mode): Likewise.
15399 (default_setup_incoming_vararg_bounds): Likewise.
15400 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
15401 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
15402 (have_whole_vector_shift): Likewise.
15403 * tree-vect-stmts.c (vectorizable_load): Likewise.
15404 * doc/tm.texi: Regenerate.
15406 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15408 Graceful degrade if Binutils PR21472 is not available.
15411 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
15412 .rodata in flash test fails.
15413 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
15414 * confgure: Regenerate.
15415 * config.in: Regenerate.
15416 * config/avr/avr.c (avr_asm_named_section)
15417 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
15418 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
15419 (avr_asm_init_sections): Same.
15421 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15423 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
15424 (fma<VH:mode>4_intrinsic): Likewise.
15425 (*fmsub<VCVTF:mode>4): Likewise.
15426 (*fmsub<VH:mode>4_intrinsic): Likewise.
15428 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15431 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
15432 Don't depend on "optimize > 0".
15433 (out_movhi_r_mr, out_movqi_mr_r): Same.
15434 (out_movhi_mr_r, out_movqi_r_mr): Same.
15435 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
15436 io_address_operand on "optimize > 0".
15438 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15440 * tree-loop-distribution.c: Add general explanantion on the pass.
15441 (generate_loops_for_partition): Mark distributed loop.
15442 (pg_add_dependence_edges): New parameter. Handle alias data
15443 dependence specially and record it in the parameter if asked.
15444 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
15445 (init_partition_graph_vertices, add_partition_graph_edge): New.
15446 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
15447 (free_partition_graph_vdata, build_partition_graph): New.
15448 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
15449 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
15450 (data_ref_segment_size, latch_dominated_by_data_ref): New.
15451 (compute_alias_check_pairs, version_loop_by_alias_check): New.
15452 (version_for_distribution_p, finalize_partitions): New.
15453 (distribute_loop): Handle alias data dependence specially. Factor
15454 out loop fusion code as functions and call these functions.
15456 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15458 * tree-loop-distribution.c (classify_partition): New parameter and
15459 better handle reduction statement.
15460 (rdg_build_partitions): Revise comment.
15461 (distribute_loop): Compute statements in all partitions and pass it
15462 to classify_partition.
15464 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15466 * tree-loop-distribution.c (enum partition_type): New.
15467 (struct partition): New field type.
15468 (partition_merge_into): Add parameter. Update partition type.
15469 (data_dep_in_cycle_p, update_type_for_merge): New functions.
15470 (build_rdg_partition_for_vertex): Compute partition type.
15471 (rdg_build_partitions): Dump partition type.
15472 (distribute_loop): Update calls to partition_merge_into.
15474 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15476 * tree-loop-distribution.c (struct ddr_hasher): New.
15477 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
15479 (classify_partition): Call get_data_dependence.
15480 (pg_add_dependence_edges): Ditto.
15481 (distribute_loop): Release data dependence hash table.
15483 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15485 * tree-loop-distribution.c (ref_base_address): Delete.
15486 (similar_memory_accesses): Rename ...
15487 (share_memory_accesses): ... to this. Check if partitions access
15488 the same memory reference.
15489 (distribute_loop): Call share_memory_accesses.
15491 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15493 * tree-loop-distribution.c (struct partition): New field recording
15494 its data reference.
15495 (partition_alloc, partition_free): Init and release data refs.
15496 (partition_merge_into): Merge data refs.
15497 (build_rdg_partition_for_vertex): Collect data refs for partition.
15498 (pg_add_dependence_edges): Change parameters from vector to bitmap.
15500 (distribute_loop): Remve data refs from vertice data of partition
15503 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15505 * tree-loop-distribution.c (params.h): Include header file.
15506 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
15507 (datarefs_vec): New global var.
15508 (create_rdg_vertices): Use datarefs_vec directly.
15509 (free_rdg): Don't free data references.
15510 (build_rdg): Update use. Don't free data references.
15511 (distribute_loop): Compute global variable for data references.
15512 Bail out if there are too many data references.
15514 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15516 * tree-loop-distribution.c (loop_nest): New global var.
15517 (build_rdg): Use loop directly, rather than loop nest.
15518 (pg_add_dependence_edges): Remove loop nest parameter. Use global
15520 (distribute_loop): Compute global variable loop nest. Update use.
15522 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15524 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
15525 (partition_merge_into): New parameter. Dump reason for fusion.
15526 (distribute_loop): Update use of partition_merge_into.
15528 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15530 * tree-loop-distribution.c (bb_top_order_index): New.
15531 (bb_top_order_index_size, bb_top_order_cmp): New.
15532 (stmts_from_loop): Use topological order.
15533 (pass_loop_distribution::execute): Compute and release topological
15534 order for basic blocks.
15536 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15538 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
15541 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15543 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
15544 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
15545 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
15546 * internal-fn.def (LOOP_DIST_ALIAS): New.
15547 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
15548 (fold_loop_internal_call): ... this.
15549 (vect_loop_dist_alias_call): New function.
15550 (set_uid_loop_bbs): Call fold_loop_internal_call.
15551 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
15554 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15557 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
15558 Require dead FLAGS_REG at the beginning of a peephole.
15560 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15563 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
15564 arguments in the call to __builtin_ia32_sbb_u32.
15565 (_subborrow_u64): Swap _X and _Y arguments in the call to
15566 __builtin_ia32_sbb_u64.
15568 2017-07-04 Jakub Jelinek <jakub@redhat.com>
15571 * tree-vrp.c (compare_assert_loc): Turn into a function template
15572 with stable template parameter. Only test if a->e is NULL,
15573 !a->e == !b->e has been verified already. Use e == NULL or
15574 e != NULL instead of e or ! e tests. If stable is true, don't use
15575 iterative_hash_expr, on the other side allow a or b or both NULL
15576 and sort the NULLs last.
15577 (process_assert_insertions): Sort using compare_assert_loc<false>
15578 instead of compare_assert_loc, later sort using
15579 compare_assert_loc<true> before calling process_assert_insertions_for
15580 in a loop. Use break instead of continue once seen NULL pointer.
15582 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15584 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15585 Cortex-R7 and Cortex-R8 processors.
15587 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15589 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
15590 uninitialized while src is not.
15592 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
15594 * common/config/arm/arm-common.c: Adjust include path for
15596 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
15597 (arm-cpu.h): Create in build directory. Adjust dependency rules.
15598 (arm-cpu-data.h): Likewise.
15599 (arm-cpu-cdata.h): Likewise.
15600 * config/arm/arm-cpu.h: Delete.
15601 * config/arm/arm-cpu-cdata.h: Delete.
15602 * config/arm/arm-cpu-data.h: Delete.
15604 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
15606 * config/arm/arm-cpus.in (cortex-a55): New.
15607 (cortex-a75): Likewise.
15608 (cortex-a75.cortex-a55): Likewise.
15609 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
15611 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
15612 * config/arm/arm-cpu-cdata.h: Regenerate.
15613 * config/arm/arm-cpu-data.h: Regenerate.
15614 * config/arm/arm-cpu.h: Regenerate.
15615 * config/arm/arm-tables.opt: Regenerate.
15616 * config/arm/arm-tune.md: Regenerate.
15618 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15620 * haifa-sched.c (sched_create_recovery_edges): Update profile.
15622 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15624 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
15627 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
15629 PR tree-optimization/81292
15630 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
15631 full_string_p, also call adjust_related_strinfos if the adjustment
15632 is simple, otherwise invalidate related strinfos.
15634 2017-07-04 Martin Liska <mliska@suse.cz>
15637 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
15638 newly created variable as DECL_IGNORED_P.
15640 2017-07-04 Martin Liska <mliska@suse.cz>
15643 * ipa-inline.c (inline_small_functions):
15644 Use xstrdup_for_dump.
15646 2017-07-04 Tom de Vries <tom@codesourcery.com>
15648 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
15650 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15653 * config/darwin.c (darwin_function_switched_text_sections):
15656 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
15658 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
15660 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15662 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
15664 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15666 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
15667 min_profitable_iters, and th as inclusive lower bounds.
15668 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
15669 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
15670 for min_profitable_iters and min_profitable_estimate.
15671 (vect_transform_loop): Treat th as an inclusive lower bound.
15672 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15674 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15677 * config/darwin.c (darwin_function_switched_text_sections):
15678 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
15679 in two pieces, and suppress the use of buf.
15681 2017-07-03 Nathan Sidwell <nathan@acm.org>
15683 * hash-table.h (hash_table_mod1): Fix indentation.
15685 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15687 PR middle-end/81290
15688 * predict.c (force_edge_cold): Be more careful about propagation
15690 * profile-count.h (profile_probability::guessed,
15691 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
15693 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
15695 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
15697 * doc/invoke.texi (rcpc architecture extension): Document it.
15699 2017-07-03 Richard Biener <rguenther@suse.de>
15701 PR tree-optimization/60510
15702 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
15703 the scalar reduction PHI and use it.
15704 (vectorizable_reduction): Properly guard the single_defuse_cycle
15705 path for non-SLP reduction chains where we cannot use it.
15706 Rework reduc_def/index and vector type deduction. Rework
15707 vector operand gathering during reduction op code-gen.
15708 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
15709 chains dissolve the chain and leave it to non-SLP reduction
15712 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15714 * tree-data-ref.h (dr_alignment): Declare.
15715 * tree-data-ref.c (dr_alignment): New function.
15716 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
15717 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
15719 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
15721 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15723 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
15724 and base_misalignment fields.
15725 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
15726 * tree-data-ref.c: Include builtins.h.
15727 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
15728 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
15729 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
15730 * tree-vect-data-refs.c: Include tree-cfg.h.
15731 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
15732 fields instead of calculating an alignment here.
15733 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
15734 innermost_loop_behavior fields.
15736 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15738 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
15740 (DR_STEP_ALIGNMENT): New macro.
15741 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
15742 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
15743 (create_data_ref): Print it.
15744 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
15745 to tell whether the step preserves vector (mis)alignment.
15746 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15747 Move the check for an integer step and generalise to all INTEGER_CST.
15748 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
15749 Print the outer step alignment.
15751 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15753 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
15754 with offset_alignment.
15755 (DR_ALIGNED_TO): Delete.
15756 (DR_OFFSET_ALIGNMENT): New macro.
15757 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
15758 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
15759 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
15760 (create_data_ref): Likewise.
15761 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15762 (vect_analyze_data_refs): Likewise.
15763 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
15764 creating dummy innermost behavior.
15766 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15768 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
15769 with a "innermost_loop_behavior *" and refeence tree.
15770 * tree-data-ref.c (dr_analyze_innermost): Likewise.
15771 (create_data_ref): Update call accordingly.
15772 * tree-predcom.c (find_looparound_phi): Likewise.
15774 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15776 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
15777 fields with dr_wrt_vec_loop.
15778 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
15779 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
15780 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
15781 (vect_dr_behavior): New function.
15782 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15783 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
15784 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
15785 track whether the step preserves the misalignment.
15786 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15787 Use vect_dr_behavior.
15788 (vect_setup_realignment): Update call accordingly.
15789 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
15790 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
15791 call to vect_create_addr_base_for_vector_ref.
15792 (vect_create_cond_for_align_checks): Likewise.
15793 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
15794 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
15795 (vect_recog_mask_conversion_pattern): Likewise.
15796 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
15797 (new_stmt_vec_info): Remove redundant zeroing.
15799 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15801 * common/config/arm/arm-common.c (arm_be8_option): New function.
15802 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
15803 (ISA_ARMv6): Add isa_bit_be8.
15804 * config/arm/arm.h (arm_be8_option): Add prototype.
15805 (BE8_SPEC_FUNCTION): New define.
15806 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
15807 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
15808 (mlittle-endian): Similarly.
15809 (mbe8, mbe32): New options.
15810 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
15811 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
15813 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15815 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
15817 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15819 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
15820 (cleanup_tree_cfg_bb): Use it.
15821 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
15823 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
15825 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15828 * loop-doloop.c (add_test): Update profile.
15830 2017-07-03 Martin Liska <mliska@suse.cz>
15833 * sanopt.c (rewrite_usage_of_param): New function.
15834 (sanitize_rewrite_addressable_params): Likewise.
15835 (pass_sanopt::execute): Call rewrite_usage_of_param.
15837 2017-07-03 Richard Biener <rguenther@suse.de>
15839 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
15840 back to using VIEW_CONVERT_EXPR.
15842 2017-07-03 Martin Liska <mliska@suse.cz>
15845 * doc/extend.texi: Document when a resolver function is
15846 generated for target_clones.
15848 2017-07-03 Martin Liska <mliska@suse.cz>
15850 * asan.c (asan_emit_stack_protection): Unpoison just red zones
15851 and shadow memory of auto variables which are subject of
15852 use-after-scope sanitization.
15853 (asan_expand_mark_ifn): Add do set only when is_poison.
15855 2016-07-03 Richard Biener <rguenther@suse.de>
15857 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
15859 (vect_force_simple_reduction): Record reduction def -> phi mapping.
15860 (vectorizable_reduction): Perform reduction PHI creation when
15861 visiting a reduction PHI and adjust and simplify code generation
15862 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
15863 (vect_transform_loop): Visit reduction PHIs.
15864 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
15865 defs into the SLP tree.
15866 (vect_build_slp_tree): Reduction defs terminate the recursion.
15867 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
15869 (vect_get_vec_defs_for_stmt_copy): Export.
15870 (vect_get_vec_defs): Likewise.
15871 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
15873 (vect_get_vec_defs_for_stmt_copy): Declare.
15874 (vect_get_vec_defs): Likewise.
15876 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15878 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
15879 parameter with a "loop" parameter and use it instead of the
15880 loop containing DR_STMT. Don't check simple_iv when doing
15881 BB analysis. Describe the two analysis modes in the comment.
15883 2017-07-03 Tom de Vries <tom@codesourcery.com>
15885 PR tree-optimization/69468
15886 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
15887 (find_same_succ_bb): Handle ignore_edge_flags.
15889 2017-07-03 Tom de Vries <tom@codesourcery.com>
15891 PR tree-optimization/81192
15892 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
15894 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
15896 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
15898 2017-07-03 Tom de Vries <tom@codesourcery.com>
15900 PR tree-optimization/81192
15901 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
15902 BB_SAME_SUCC (bb) == NULL.
15904 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15906 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
15909 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15911 * dumpfile.c: Include profile-count.h
15912 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
15914 (insert_cond_bb): Update profile.
15915 * tree-cfg.h (insert_cond_bb): Update prototype.
15916 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
15917 * tree-dump.c: Do not include tree-cfg.
15919 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15921 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
15923 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15925 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
15928 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15930 * tree-complex.c (expand_complex_div_wide): update profile.
15932 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15933 Alan Hayward <alan.hayward@arm.com>
15934 David Sherwood <david.sherwood@arm.com>
15936 * Makefile.in (MACHMODE_H): Remove insn-modes.h
15937 (CORETYPES_H): New define.
15938 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
15939 (insn-modes-inline.h, s-modes-inline-h): New rules.
15940 (generated_files): Add insn-modes-inline.h.
15941 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
15942 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
15943 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
15944 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
15945 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
15946 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
15947 (build/gencodes.o, build/genconditions.o): Likewise.
15948 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
15949 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
15950 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
15951 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
15952 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
15953 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
15954 * coretypes.h: Include everything up to real.h for generators.
15955 Include insn-modes.h first. Include wide-int-print.h after
15956 wide-int.h. Include insn-modes-inline.h and then machmode.h.
15957 * machmode.h: Don't include insn-modes.h here.
15958 * function-tests.c: Remove includes of signop.h, machmode.h,
15959 double-int.h and wide-int.h.
15961 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
15963 * optc-save-gen.awk: Likewise.
15964 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
15965 * godump.c: Remove include of wide-int-print.h.
15966 * pretty-print.h: Likewise.
15967 * wide-int-print.cc: Likewise.
15968 * wide-int.cc: Likewise.
15969 * hash-map-tests.c: Remove include of signop.h.
15970 * hash-set-tests.c: Likewise.
15971 * rtl-tests.c: Likewise.
15972 * mkconfig.sh: Remove include of machmode.h.
15973 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
15975 (emit_insn_modes_inline_h): ...this new function. Emit the code
15976 into an insn-modes-inline.h header file, adding appropriate
15977 include guards and end comments.
15978 (emit_insn_modes_c_header): Remove include of machmode.h.
15979 (emit_min_insn_modes_c_header): Include coretypes.h rather than
15981 (main): Handle -i flag and call emit_insn_modes_inline_h when
15984 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15986 * tree-ssa-strlen.c (strinfo): Rename the length field to
15987 nonzero_chars. Add a full_string_p field.
15988 (compare_nonzero_chars, zero_length_string_p): New functions.
15989 (get_addr_stridx): Add an offset_out parameter.
15990 Use compare_nonzero_chars.
15991 (get_stridx): Update accordingly. Use compare_nonzero_chars.
15992 (new_strinfo): Update after above changes to strinfo.
15993 (set_endptr_and_length): Set full_string_p.
15994 (get_string_length): Update after above changes to strinfo.
15995 (unshare_strinfo): Update call to new_strinfo.
15996 (maybe_invalidate): Likewise.
15997 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
15998 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
15999 as a uhwi instead of an shwi. Update after above changes to
16000 strinfo and new_strinfo.
16001 (zero_length_string): Assert that chainsi contains full strings.
16002 Use zero_length_string_p. Update call to new_strinfo.
16003 (adjust_related_strinfos): Update after above changes to strinfo.
16004 Copy full_string_p from origsi.
16005 (adjust_last_stmt): Use zero_length_string_p.
16006 (handle_builtin_strlen): Update after above changes to strinfo and
16007 new_strinfo. Install the lhs as the string length if the previous
16008 entry didn't describe a full string.
16009 (handle_builtin_strchr): Update after above changes to strinfo
16011 (handle_builtin_strcpy): Likewise.
16012 (handle_builtin_strcat): Likewise.
16013 (handle_builtin_malloc): Likewise.
16014 (handle_pointer_plus): Likewise.
16015 (handle_builtin_memcpy): Likewise. Track nonzero characters
16016 that aren't necessarily followed by a nul terminator.
16017 (handle_char_store): Likewise.
16019 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16021 PR tree-optimization/80769
16022 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
16023 for malloc and calloc. Document the new invariant that all related
16024 strinfos have delayed lengths or none do.
16025 (verify_related_strinfos): Move earlier in file.
16026 (set_endptr_and_length): New function, split out from...
16027 (get_string_length): ...here. Also set the lengths of related
16029 (zero_length_string): Assert that chainsi has known (rather than
16031 (adjust_related_strinfos): Likewise.
16033 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16035 PR tree-optimization/81136
16036 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
16037 assert that two references with the same misalignment have the same
16038 compile-time misalignment if those compile-time misalignments
16041 2017-07-01 Andi Kleen <ak@linux.intel.com>
16043 * print-tree.c (print_node): Print all attributes.
16045 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16047 * cfg.c (scale_bbs_frequencies): New function.
16048 * cfg.h (scale_bbs_frequencies): Declare it.
16049 * cfgloopanal.c (single_likely_exit): Cleanup.
16050 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
16052 (scale_loop_profile): Likewise.
16053 (loop_version): Likewise.
16054 (create_empty_loop_on_edge): Update.
16055 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
16056 scale_loop_frequencies, scale_loop_profile, loopify,
16057 loop_version): Update prototypes.
16058 * modulo-sched.c (sms_schedule): Update.
16059 * predict.c (unlikely_executed_edge_p): Also check probability.
16060 (probably_never_executed_edge_p): Fix typo.
16061 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16062 * tree-parloops.c (gen_parallel_loop): Update.
16063 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
16064 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16065 * tree-ssa-loop-split.c (split_loop): Update.
16066 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16067 * tree-vect-loop-manip.c (vect_do_peeling): Update.
16068 (vect_loop_versioning): Update.
16069 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16071 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16073 * trans-mem.c (split_bb_make_tm_edge): Update profile.
16075 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16077 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
16078 to keep profile consistent.
16080 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16082 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
16083 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
16084 * profile-count.h (max_safe_multiplier): Make unsigned.
16085 (profile_count::guessed_zero): New.
16087 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16089 * bb-reorder.c (fix_up_crossing_landing_pad,
16090 fix_crossing_conditional_branches): Use make_single_succ_edge
16091 to keep profile consistent.
16093 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16095 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
16098 2017-07-01 Jakub Jelinek <jakub@redhat.com>
16101 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
16102 the right scopes, make sure cond_jump isn't preserved between multiple
16103 iterations. Search for fallthru edge whenever there are 3+ edges and
16104 use find_fallthru_edge for it.
16106 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16108 Patch by Alexander Monakov <amonakov@ispras.ru>
16109 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
16110 probabilities consistently.
16112 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16114 * pa.c (pa_expand_compare_and_swap_loop): Update call of
16115 emit_cmp_and_jump_insns.
16117 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16120 * tree-inline.c (expand_call_inline): Combine profile statuses.
16122 2017-06-30 Andrew Pinski <apinski@cavium.com>
16124 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
16125 fold_stmt returned true.
16127 2017-06-30 Nathan Sidwell <nathan@acm.org>
16129 * ggc.h (empty_string): Delete.
16130 * cfgexpand.c (expand_asm_stmt): Use plain "".
16131 * optabs.c (expand_asm_memory_barrier): Likewise.
16132 * stringpool.c (empty_string): Delete.
16133 (digit_vector, digit_string): Delete.
16134 (ggc_alloc_string): Use plain "", don't optimize single digit
16135 strings. Use ggc_alloc_atomic.
16137 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
16139 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
16140 comparison set and one other set, use the cost of the non-comparison
16143 2017-06-30 Nathan Sidwell <nathan@acm.org>
16145 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
16148 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
16150 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
16151 loops. Remove now unneeded calls to gimple_switch_set_label() that
16152 just set removed labels to NULL_TREE.
16154 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
16156 * tree-ssanames.c (set_range_info_raw): Abstract from ...
16157 (set_range_info): ...here. Only call set_range_info_raw if domain
16159 (set_nonzero_bits): Call set_range_info_raw.
16160 * tree-ssanames.h (set_range_info_raw): New.
16162 2017-06-30 Jakub Jelinek <jakub@redhat.com>
16165 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
16166 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
16167 of nonimmediate_operand and <store_mask_constraint> instead of m
16168 for the input operand. For V8FI iterator, always split if input
16169 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
16170 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
16171 <store_mask_predicate> instead of register_operand and
16172 <store_mask_constraint> instead of v for the input operand. Make
16173 sure both operands aren't MEMs for if not <mask_applied>.
16175 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
16177 * lto-wrapper.c (copy_file) Close both file descriptors before
16180 2017-06-30 Martin Liska <mliska@suse.cz>
16183 * multiple_target.c (create_dispatcher_calls): Make ifunc
16184 also for function that don't have calls or are not referenced.
16186 2017-06-30 Richard Biener <rguenther@suse.de>
16188 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
16189 analyze the first scalar stmt. Move vector type computation
16190 for the BB case here from ...
16191 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
16192 live operation processing in the SLP case properly.
16194 2017-06-30 Richard Biener <rguenther@suse.de>
16196 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
16198 2017-06-30 Martin Liska <mliska@suse.cz>
16201 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
16202 before BUILT_IN_UNWIND_RESUME when ASAN is used.
16204 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
16206 * doc/invoke.texi (AArch64): Add missing options and remove redundant
16209 2017-06-30 Richard Biener <rguenther@suse.de>
16211 PR tree-optimization/81249
16212 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
16213 condition reduction result to original scalar type.
16215 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16217 * profile-count.h (enum profile_quality): Fix typos and whitespace
16220 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16222 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
16223 type for branch probabilities.
16225 2017-06-29 Julian Brown <julian@codesourcery.com>
16226 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16228 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
16229 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
16230 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
16231 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
16233 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16235 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
16236 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
16237 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
16238 CC usage from generic code to here.
16239 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
16240 CC usage into the target macros.
16242 2017-06-29 Maya Rashish <coypu@sdf.org>
16244 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
16247 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16249 * arm/arm-builtins.c: Include profile-count.h
16250 * except.c (sjlj_emit_function_enter): Use
16251 profile_probability::unlikely.
16253 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16255 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
16256 and tocrel_offset be pointer args rather than implicitly using
16258 (legitimate_constant_pool_address_p, rs6000_emit_move,
16259 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
16260 tocrel_offset and use in toc_relative_expr_p call.
16261 (print_operand, print_operand_address): Use static tocrel_base_oac
16262 and tocrel_offset_oac.
16263 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
16266 2017-06-29 Maya Rashish <coypu@sdf.org>
16268 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
16270 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
16272 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
16273 objects, take into account only the alignment of 'op0' and 'mode1' if
16276 2017-06-29 Steve Ellcey <sellcey@cavium.com>
16278 * ccmp.c (ccmp_tree_comparison_p): New function.
16279 (ccmp_candidate_p): Update to use above function.
16280 (get_compare_parts): New function.
16281 (expand_ccmp_next): Update to use new functions.
16282 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
16284 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
16285 take mode as argument.
16286 * ccmp.h (expand_ccmp_expr): Add mode as argument.
16287 * expr.c (expand_expr_real_1): Pass mode as argument.
16289 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
16291 * combine.c (combine_instructions): Print insns to dump_file, together
16294 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16296 * asan.c (asan_emit_stack_protection): Update.
16297 (create_cond_insert_point): Update.
16298 * auto-profile.c (afdo_propagate_circuit): Update.
16299 * basic-block.h (struct edge_def): Turn probability to
16300 profile_probability.
16301 (EDGE_FREQUENCY): Update.
16302 * bb-reorder.c (find_traces_1_round): Update.
16303 (better_edge_p): Update.
16304 (sanitize_hot_paths): Update.
16305 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
16306 (make_single_succ_edge): Update.
16307 (check_bb_profile): Update.
16308 (dump_edge_info): Update.
16309 (update_bb_profile_for_threading): Update.
16310 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
16311 probabilitycount to 0.
16312 * cfgbuild.c (compute_outgoing_frequencies): Update.
16313 * cfgcleanup.c (try_forward_edges): Update.
16314 (outgoing_edges_match): Update.
16315 (try_crossjump_to_edge): Update.
16316 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
16317 (expand_gimple_tailcall): Update.
16318 (construct_init_block): Use make_single_succ_edge.
16319 (construct_exit_block): Use make_single_succ_edge.
16320 * cfghooks.c (verify_flow_info): Update.
16321 (redirect_edge_succ_nodup): Update.
16322 (split_edge): Update.
16323 (account_profile_record): Update.
16324 * cfgloopanal.c (single_likely_exit): Update.
16325 * cfgloopmanip.c (scale_loop_profile): Update.
16326 (set_zero_probability): Remove.
16327 (duplicate_loop_to_header_edge): Update.
16328 * cfgloopmanip.h (loop_version): Update prototype.
16329 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
16330 (force_nonfallthru_and_redirect): Update.
16331 (update_br_prob_note): Update.
16332 (rtl_verify_edges): Update.
16333 (purge_dead_edges): Update.
16334 (rtl_lv_add_condition_to_bb): Update.
16335 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
16336 * cgraphunit.c (init_lowered_empty_function): Update.
16337 (cgraph_node::expand_thunk): Update.
16338 * cilk-common.c: Include profile-count.h
16339 * dojump.c (inv): Remove.
16340 (jumpifnot): Update.
16341 (jumpifnot_1): Update.
16342 (do_jump_1): Update.
16344 (do_jump_by_parts_greater_rtx): Update.
16345 (do_compare_rtx_and_jump): Update.
16346 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
16347 do_jump_1. do_compare_rtx_and_jump): Update prototype.
16348 * dwarf2cfi.c: Include profile-count.h
16349 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
16350 (sjlj_emit_dispatch_table): Likewise.
16351 * explow.c: Include profile-count.h
16352 * expmed.c (emit_store_flag_force): Update.
16353 (do_cmp_and_jump): Update.
16354 * expr.c (compare_by_pieces_d::generate): Update.
16355 (compare_by_pieces_d::finish_mode): Update.
16356 (emit_block_move_via_loop): Update.
16357 (store_expr_with_bounds): Update.
16358 (store_constructor): Update.
16359 (expand_expr_real_2): Update.
16360 (expand_expr_real_1): Update.
16361 * expr.h (try_casesi, try_tablejump): Update prototypes.
16362 * gimple-pretty-print.c (dump_probability): Update.
16363 (dump_profile): New.
16364 (dump_gimple_label): Update.
16365 (dump_gimple_bb_header): Update.
16366 * graph.c (draw_cfg_node_succ_edges): Update.
16367 * hsa-gen.c (convert_switch_statements): Update.
16368 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
16369 (find_if_case_1): Update.
16370 (find_if_case_2): Update.
16371 * internal-fn.c (expand_arith_overflow_result_store): Update.
16372 (expand_addsub_overflow): Update.
16373 (expand_neg_overflow): Update.
16374 (expand_mul_overflow): Update.
16375 (expand_vector_ubsan_overflow): Update.
16376 * ipa-cp.c (good_cloning_opportunity_p): Update.
16377 * ipa-split.c (split_function): Use make_single_succ_edge.
16378 * ipa-utils.c (ipa_merge_profiles): Update.
16379 * loop-doloop.c (add_test): Update.
16380 (doloop_modify): Update.
16381 * loop-unroll.c (compare_and_jump_seq): Update.
16382 (unroll_loop_runtime_iterations): Update.
16383 * lra-constraints.c (lra_inheritance): Update.
16384 * lto-streamer-in.c (input_cfg): Update.
16385 * lto-streamer-out.c (output_cfg): Update.
16386 * mcf.c (adjust_cfg_counts): Update.
16387 * modulo-sched.c (sms_schedule): Update.
16388 * omp-expand.c (expand_omp_for_init_counts): Update.
16389 (extract_omp_for_update_vars): Update.
16390 (expand_omp_ordered_sink): Update.
16391 (expand_omp_for_ordered_loops): Update.
16392 (expand_omp_for_generic): Update.
16393 (expand_omp_for_static_nochunk): Update.
16394 (expand_omp_for_static_chunk): Update.
16395 (expand_cilk_for): Update.
16396 (expand_omp_simd): Update.
16397 (expand_omp_taskloop_for_outer): Update.
16398 (expand_omp_taskloop_for_inner): Update.
16399 * omp-simd-clone.c (simd_clone_adjust): Update.
16400 * optabs.c (expand_doubleword_shift): Update.
16401 (expand_abs): Update.
16402 (emit_cmp_and_jump_insn_1): Update.
16403 (expand_compare_and_swap_loop): Update.
16404 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
16405 * predict.c (predictable_edge_p): Update.
16406 (edge_probability_reliable_p): Update.
16407 (set_even_probabilities): Update.
16408 (combine_predictions_for_insn): Update.
16409 (combine_predictions_for_bb): Update.
16410 (propagate_freq): Update.
16411 (estimate_bb_frequencies): Update.
16412 (force_edge_cold): Update.
16413 * profile-count.c (profile_count::dump): Add missing space into dump.
16414 (profile_count::debug): Add newline.
16415 (profile_count::differs_from_p): Explicitly convert to unsigned.
16416 (profile_count::stream_in): Update.
16417 (profile_probability::dump): New member function.
16418 (profile_probability::debug): New member function.
16419 (profile_probability::differs_from_p): New member function.
16420 (profile_probability::differs_lot_from_p): New member function.
16421 (profile_probability::stream_in): New member function.
16422 (profile_probability::stream_out): New member function.
16423 * profile-count.h (profile_count_quality): Rename to ...
16424 (profile_quality): ... this one.
16425 (profile_probability): New.
16426 (profile_count): Update.
16427 * profile.c (compute_branch_probabilities): Update.
16428 * recog.c (peep2_attempt): Update.
16429 * sched-ebb.c (schedule_ebbs): Update.
16430 * sched-rgn.c (find_single_block_region): Update.
16431 (compute_dom_prob_ps): Update.
16432 (schedule_region): Update.
16433 * sel-sched-ir.c (compute_succs_info): Update.
16434 * stmt.c (struct case_node): Update.
16435 (do_jump_if_equal): Update.
16436 (get_outgoing_edge_probs): Update.
16437 (conditional_probability): Update.
16438 (emit_case_dispatch_table): Update.
16439 (expand_case): Update.
16440 (expand_sjlj_dispatch_table): Update.
16441 (emit_case_nodes): Update.
16442 * targhooks.c: Update.
16443 * tracer.c (better_p): Update.
16444 (find_best_successor): Update.
16445 * trans-mem.c (expand_transaction): Update.
16446 * tree-call-cdce.c: Update.
16447 * tree-cfg.c (gimple_split_edge): Upate.
16448 (move_sese_region_to_fn): Upate.
16449 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
16450 * tree-eh.c (lower_resx): Upate.
16451 (cleanup_empty_eh_move_lp): Upate.
16452 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16453 * tree-inline.c (copy_edges_for_bb): Update.
16454 (copy_cfg_body): Update.
16455 * tree-parloops.c (gen_parallel_loop): Update.
16456 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
16457 (gimple_gen_time_profiler): Update.
16458 * tree-ssa-dce.c (remove_dead_stmt): Update.
16459 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16460 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
16461 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
16462 (unloop_loops): Update.
16463 (try_peel_loop): Update.
16464 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16465 * tree-ssa-loop-split.c (connect_loops): Update.
16466 (split_loop): Update.
16467 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16468 (hoist_guard): Update.
16469 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
16470 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
16471 (value_replacement): Update.
16472 * tree-ssa-reassoc.c (branch_fixup): Update.
16473 * tree-ssa-tail-merge.c (replace_block_by): Update.
16474 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
16475 (create_edge_and_update_destination_phis): Update.
16476 (compute_path_counts): Update.
16477 (recompute_probabilities): Update.
16478 (update_joiner_offpath_counts): Update.
16479 (freqs_to_counts_path): Update.
16480 (duplicate_thread_path): Update.
16481 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
16482 (struct switch_conv_info): Update.
16483 (gen_inbound_check): Update.
16484 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16485 (vect_do_peeling): Update.
16486 (vect_loop_versioning): Update.
16487 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16488 (optimize_mask_stores): Update.
16489 * ubsan.c (ubsan_expand_null_ifn): Update.
16490 * value-prof.c (gimple_divmod_fixed_value): Update.
16491 (gimple_divmod_fixed_value_transform): Update.
16492 (gimple_mod_pow2): Update.
16493 (gimple_mod_pow2_value_transform): Update.
16494 (gimple_mod_subtract): Update.
16495 (gimple_mod_subtract_transform): Update.
16496 (gimple_ic): Update.
16497 (gimple_stringop_fixed_value): Update.
16498 (gimple_stringops_transform): Update.
16499 * value-prof.h: Update.
16501 2017-06-29 Carl Love <cel@us.ibm.com>
16503 * config/rs6000/rs6000-c.c: Add support for built-in functions
16504 vector signed int vec_signed (vector float);
16505 vector signed long long vec_signed (vector double);
16506 vector signed int vec_signed2 (vector double, vector double);
16507 vector signed int vec_signede (vector double);
16508 vector signed int vec_signedo (vector double);
16509 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
16510 instruction generator.
16511 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16512 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
16513 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
16515 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
16516 vunsignede_v2df): Add define_expands.
16517 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
16518 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
16519 VEC_UNSIGNEDO): Add definitions.
16520 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16521 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
16522 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
16523 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
16524 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
16525 * config/rs6000/altivec.h (vec_signed, vec_signed2,
16526 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
16527 vec_unsignede, vec_unsignedo): Add builtin defines.
16528 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
16530 * doc/extend.texi: Update the built-in documentation file for the
16531 new built-in functions.
16533 2017-06-29 Richard Biener <rguenther@suse.de>
16535 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
16536 reduction chains to LOOP_VINFO_REDUCTIONS.
16537 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
16538 SLP reductions after processing reduction chains.
16540 2017-06-29 Nathan Sidwell <nathan@acm.org>
16542 * builtins.c (fold_builtin_FUNCTION): Use
16543 lang_hooks.decl_printable_name.
16545 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
16547 PR middle-end/81194
16548 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
16549 with only one label.
16550 * stmt.c (expand_case): Assert NCASES is greater than one.
16552 2017-06-29 Richard Biener <rguenther@suse.de>
16554 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
16556 (group_case_labels): Likewise.
16557 (find_taken_edge): Push sanity checking on val to workers...
16558 (find_taken_edge_cond_expr): ... here
16559 (find_taken_edge_switch_expr): ... and here, handle cases
16560 with just a default label.
16561 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
16562 (group_case_labels): Likewise.
16563 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
16564 group_case_labels does anything cleanup the CFG again.
16566 2017-06-29 Bin Cheng <bin.cheng@arm.com>
16568 PR tree-optimization/81196
16569 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
16570 exit condition comparing two IVs.
16572 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
16574 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
16575 profile to the dummy entry at the end of the list of architectures.
16576 * config/arm/arm-cpu-cdata.h: Regenerated.
16578 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16579 Michael Collison <michael.collison@arm.com>
16582 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
16584 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
16585 (*aarch64_reg_<mode>3_minus_mask): New pattern.
16586 (*aarch64_<optab>_reg_di3_mask2): New pattern.
16587 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
16588 of shift when the shift amount is masked with constant equal to
16589 the size of the mode.
16590 * config/aarch64/predicates.md (subreg_lowpart_operator): New
16593 2017-06-29 Martin Liska <mliska@suse.cz>
16595 * config/i386/i386.opt: Change range from [1,5] to [0,5].
16597 2017-06-29 Yury Gribov <tetra2005@gmail.com>
16600 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
16603 (edge_growth_cache_entry::edge_growth_cache_entry): New
16605 (reset_edge_growth_cache): Update to use constructor.
16607 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16609 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
16610 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16611 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16613 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
16615 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
16616 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
16618 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
16620 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
16621 (*-linux-uclibc*): Add t-uclibc tmake_file.
16622 * config/t-musl: New.
16623 * config/t-uclibc: New.
16625 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
16627 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
16629 (gen_comm_data): Emit architectural setting of arch_prof.
16630 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
16632 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
16633 (armv8-m.base, armv8-m.main): Likewise.
16634 * arm-protos.h (arm_build_target): Add profile field.
16635 (arch_option): Likewise.
16636 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
16638 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
16639 arm_active_target.profile.
16641 2017-06-28 Richard Biener <rguenther@suse.de>
16643 PR middle-end/81227
16644 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
16645 TYPE_OVERFLOW_WRAPS.
16646 * match.pd (negate_expr_p): Likewise.
16647 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
16648 fold_build2, not fold_binary.
16650 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16652 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
16653 Convert memory address to Pmode.
16654 (aarch64_print_operand): Assert MEM operands are always Pmode.
16656 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16659 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
16660 Remove redundant if.
16661 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
16662 * config/arm/aarch-common-protos.h
16663 (aarch_forward_to_shift_is_not_shifted_re): Remove.
16664 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
16666 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
16669 * multiple_target.c (create_dispatcher_calls): Set the default
16670 clone to be static, not public.
16672 2017-06-28 Richard Biener <rguenther@suse.de>
16674 * tree-vect-loop.c (vectorizable_reduction): Move special
16675 cond reduction IV var creation ...
16676 (vect_create_epilog_for_reduction): ... here. Remove induction_index
16677 parameter. Use STMT_VINFO_VECTYPE.
16678 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
16681 2017-06-28 Martin Liska <mliska@suse.cz>
16684 * ipa-visibility.c (non_local_p): Handle visibility.
16686 2017-06-28 Martin Liska <mliska@suse.cz>
16689 * common.opt: Add IntegerRange to various options.
16690 * opt-functions.awk (integer_range_info): New function.
16691 * optc-gen.awk: Add integer_range_info to cl_options struct.
16692 * opts-common.c (decode_cmdline_option): Handle
16693 CL_ERR_INT_RANGE_ARG.
16694 (cmdline_handle_error): Likewise.
16695 * opts.c (print_filtered_help): Show valid interval in
16696 when --help is provided.
16697 * opts.h (struct cl_option): Add range_min and range_max fields.
16698 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
16700 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
16702 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
16703 (x * C EQ/NE y * C): New transformation.
16705 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
16707 * genmultilib (combination_space): Accept '+' in option names.
16709 2017-06-28 Martin Liska <mliska@suse.cz>
16712 * asan.c (instrument_derefs): Bail out inner references
16713 that are hard register variables.
16715 2017-06-28 Jakub Jelinek <jakub@redhat.com>
16718 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
16719 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
16721 2017-06-28 Richard Biener <rguenther@suse.de>
16723 * tree-vectorizer.h (vect_get_vec_defs): Remove.
16724 (vect_get_slp_defs): Adjust.
16725 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
16727 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
16729 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
16730 get_initial_defs_for_reduction instead of vect_get_vec_defs.
16731 (vectorizable_reduction): Adjust.
16732 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
16734 (vect_get_slp_defs): Likewise.
16735 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
16736 (vectorizable_bswap): Adjust.
16737 (vectorizable_call): Likewise.
16738 (vectorizable_conversion): Likewise.
16739 (vectorizable_assignment): Likewise.
16740 (vectorizable_shift): Likewise.
16741 (vectorizable_operation): Likewise.
16742 (vectorizable_store): Likewise.
16743 (vectorizable_condition): Likewise.
16744 (vectorizable_comparison): Likewise.
16746 2017-06-28 Michael Collison <michael.collison@arm.com>
16749 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
16751 (arm_gen_movmemqi): Removed unused variable 'i'.
16752 Convert 'for' loop into 'while' loop.
16753 (arm_expand_prologue): Remove last unnecessary set of insn.
16754 (thumb_pop): Remove unused variable 'pushed_words'.
16755 (thumb_exit): Remove last unnecessary set of regs_to_pop.
16757 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16759 * config/s390/predicates.md: Use s390_rel_address_ok_p.
16760 * config/s390/s390-protos.h: Add prototype of
16761 s390_rel_address_ok_p.
16762 * config/s390/s390.c (s390_got_symbol): New function.
16763 (s390_rel_address_ok_p): New function.
16764 (legitimize_pic_address): Use s390_rel_address_ok_p.
16765 (s390_load_got): Use s390_got_symbol.
16766 (s390_option_override): Issue error if
16767 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
16768 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
16770 * config/s390/s390.opt: New option mpic-data-is-text-relative.
16772 2017-06-27 Andrew Pinski <apinski@cavium.com>
16774 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
16775 (X * copysign (1.0, X)): New pattern.
16776 (X * copysign (1.0, -X)): New pattern.
16777 (copysign (-1.0, CST)): New pattern.
16779 2017-06-27 Joseph Myers <joseph@codesourcery.com>
16781 * genmultilib (combination_space): Remove variable.
16782 Validate reuse rules against regular expression for any sequence
16783 of multilib options in any order.
16785 2017-06-27 Michael Collison <michael.collison@arm.com>
16787 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
16788 call aarch64_split_simd_combine.
16789 * (aarch64_combine_internal<mode>): Delete pattern.
16790 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
16791 Allow register and subreg operands.
16793 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16795 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
16796 specific need, just fallback on defaults.
16797 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
16799 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16800 Olivier Hainque <hainque@adacore.com>
16802 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
16804 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
16805 targets. Pick a default if no particular attempt applied.
16806 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
16809 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16811 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
16812 (x86_64-wrs-vxworks7): Likewise.
16814 2017-06-27 Marek Polacek <polacek@redhat.com>
16817 * ubsan.c (instrument_null): Check get_base_address's result for null.
16819 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16821 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
16823 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16825 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
16826 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
16827 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
16828 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
16829 New function types.
16830 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
16831 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
16832 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
16833 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
16834 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
16835 BUILT_IN_FEUPDATEENV): New builtins.
16836 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
16837 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
16838 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
16839 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
16841 (builtin_structptr_types): Adjust size.
16842 * tree.c (builtin_structptr_types): Add four entries.
16844 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16845 Olivier Hainque <hainque@adacore.com>
16847 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
16848 (TLS_SYM): New local macro, forcing reference to __tls__ on
16849 link command lines for VxWorks 7 RTPs, triggering initialization
16851 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
16852 OS features TLS support, true for RTPs on VxWorks 7.
16853 * config/vxworks.c (vxworks_override_options): Setup emutls
16856 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16858 * predict.c (test_prediction_value_range): Use -1U instead of -1
16859 to avoid narrowing conversion warning.
16860 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
16861 to avoid narrowing conversion warning.
16862 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
16864 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
16866 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16868 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
16869 64bit configurations.
16870 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
16871 (SIZE_TYPE): Likewise.
16872 * config/vxworks.c (vxworks_emutls_var_fields): Use
16873 long_unsigned_type_node instead of unsigned_type_node as the offset
16874 field type, which is "pointer" mode in emutls.c.
16876 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16879 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
16881 PR middle-end/81207
16882 * gimple-fold.c (replace_call_with_call_and_fold): Handle
16883 gimple_vuse copying separately from gimple_vdef copying.
16885 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16887 * value-prof.c (free_hist): Remove call to memset and the enclosing if
16890 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16891 Olivier Hainque <hainque@adacore.com>
16893 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
16894 for all vxworks7 targets.
16895 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
16896 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
16897 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
16898 variations for VX6/VX7 and 32/64bits later on in ...
16899 (VXWORKS_LIB_SPEC): Leverage new macros.
16900 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
16901 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
16903 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16905 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
16906 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
16908 2017-06-26 Carl Love <cel@us.ibm.com>
16910 * config/rs6000/rs6000-c.c: Add support for built-in functions
16911 vector bool char vec_reve (vector bool char);
16912 vector signed char vec_reve (vector signed char);
16913 vector unsigned char vec_reve (vector unsigned char);
16914 vector bool int vec_reve (vector bool int);
16915 vector signed int vec_reve (vector signed int);
16916 vector unsigned int vec_reve (vector unsigned int);
16917 vector bool long long vec_reve (vector bool long long);
16918 vector signed long long vec_reve (vector signed long long);
16919 vector unsigned long long vec_reve (vector unsigned long long);
16920 vector bool short vec_reve (vector bool short);
16921 vector signed short vec_reve (vector signed short);
16922 vector double vec_reve (vector double);
16923 vector float vec_reve (vector float);
16924 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
16925 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
16926 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
16927 (altivec_vreve): New pattern.
16928 * config/rs6000/altivec.h (vec_reve): New define.
16929 * doc/extend.texi (vec_rev): Update the built-in documentation file
16930 for the new built-in functions.
16932 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16934 PR tree-optimization/71815
16935 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
16937 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
16939 (slsr_process_phi): Likewise.
16940 (replace_uncond_cands_and_profitable_phis): Don't replace a
16941 multiply candidate with a stride of 1 (copy or cast).
16942 (phi_incr_cost): Call uses_consumed_by_stmt rather than
16944 (lowest_cost_path): Likewise.
16945 (total_savings): Likewise.
16947 2017-06-26 Richard Biener <rguenther@suse.de>
16950 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
16951 Use def_builtin_pure for all gather builtins.
16953 2017-06-26 Richard Biener <rguenther@suse.de>
16955 PR tree-optimization/81203
16956 * tree-tailcall.c (find_tail_calls): Do not move stmts into
16957 non-dominating BBs.
16959 2017-06-26 Marek Polacek <polacek@redhat.com>
16962 * doc/invoke.texi: Document -Wmultistatement-macros.
16964 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
16966 * doc/sourcebuild.texi (ARM-specific attributes): Document new
16967 arm_neon_ok_no_float_abi effective target.
16969 2017-06-26 Richard Biener <rguenther@suse.de>
16971 PR tree-optimization/80928
16972 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
16973 (copy_bbs): Set BB_DUPLICATED flag early.
16974 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
16976 (execute_on_shrinking_pred): Likewise.
16977 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
16978 BB_DUPLICATED blocks.
16979 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
16980 iterate over all PHIs considering removal of *gsi.
16982 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
16984 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
16987 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16989 * config/rs6000/rs6000-string.c: (expand_block_clear,
16990 do_load_for_compare, select_block_compare_mode,
16991 compute_current_alignment, expand_block_compare,
16992 expand_strncmp_align_check, expand_strn_compare,
16993 expand_block_move, rs6000_output_load_multiple)
16994 Move functions related to string/block move/compare
16995 to a separate file.
16996 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
16997 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
16998 for this function which is now used in two files.
16999 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
17000 * config.gcc: Add rs6000-string.o to extra_objs for
17001 targets powerpc*-*-* and rs6000*-*-*.
17003 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
17006 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
17007 32-bit, since indexed is not valid for DImode.
17008 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
17009 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
17010 (define_peephole2 for Altivec d-form load): Add 32-bit support.
17011 (define_peephole2 for Altivec d-form store): Likewise.
17014 * multiple_target.c (create_dispatcher_calls): Only create the
17015 dispatcher call if the function is the default clone of a
17016 versioned function.
17018 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
17020 PR middle-end/80902
17021 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
17022 a call, force the call to not be a tail call.
17024 2017-06-23 Jeff Law <law@redhat.com>
17026 * doc/contrib.texi: Add entry for Steven Pemberton's work on
17029 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
17031 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
17032 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
17033 handling for early expansion of vector shifts (sl,sr,sra,rl).
17034 (builtin_function_type): Add vector shift right instructions
17035 to the unsigned argument list.
17037 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
17039 rtl-optimizatoin/79286
17040 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
17041 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
17042 trap. PIC register plus a const unspec without offset can never trap.
17044 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
17046 * tree.h (builtin_structptr_type): New type.
17047 (builtin_structptr_types): Declare new array.
17048 * tree.c (builtin_structptr_types): New array.
17049 (free_lang_data, build_common_tree_nodes): Use it.
17051 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
17054 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
17057 2017-06-22 Matt Turner <mattst88@gmail.com>
17059 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
17060 Lake models to skylake case. Assume skylake for unknown
17061 models with clflushopt.
17063 2017-06-22 Jeff Law <law@redhat.com>
17065 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
17066 frame sizes that do not satisfy aarch64_uimm12_shift.
17068 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
17070 * profile-count.h (apply_probability,
17071 apply_scale, probability_in): Fix checks for zero.
17073 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17075 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
17076 * doc/cppdiropts.texi (-I @var{dir}): Document it.
17078 2016-06-22 Richard Biener <rguenther@suse.de>
17080 * tree-vect-loop.c (vect_model_reduction_cost): Handle
17081 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
17082 REDUC_MAX_EXPR support.
17083 (vectorizable_reduction): Likewise.
17084 (vect_create_epilog_for_reduction): Likewise.
17086 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
17088 * match.pd (A / (1 << B) -> A >> B): New.
17089 * generic-match-head.c: Include optabs-tree.h.
17090 * gimple-match-head.c: Likewise.
17091 * optabs-tree.h (target_supports_op_p): New.
17092 * optabs-tree.c (target_supports_op_p): New.
17094 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17096 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
17097 $gcc_cv_ld --help output.
17098 (gcc_cv_ld_demangle): Likewise.
17099 (gcc_cv_ld_eh_frame_hdr): Likewise.
17100 (gcc_cv_ld_pie): Likewise.
17101 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
17102 (gcc_cv_ld_buildid): Likewise.
17103 (gcc_cv_ld_sysroot): Likewise.
17104 (ld_bndplt_support): Likewise.
17105 (ld_pushpopstate_support): Likewise.
17106 * configure: Regenerate.
17107 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
17109 2017-06-21 Jakub Jelinek <jakub@redhat.com>
17112 * config/i386/sse.md (round<mode>2): Renumber match_dup and
17113 operands indexes to avoid gap between operands and match_dups.
17115 2017-06-21 Andrew Pinski <apinski@cavium.com>
17117 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
17118 Increment Arith_shift and Arith_shift_reg by 1.
17119 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
17121 * config/aarch64/aarch64.c (thunderx_tunings): Enable
17122 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
17123 (aarch64_strip_extend): Add new argument and test for it.
17124 (aarch64_cheap_mult_shift_p): New function.
17125 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
17126 add a cost if it is true.
17127 Update calls to aarch64_strip_extend.
17128 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
17130 2017-06-21 Andrew Pinski <apinski@cavium.com>
17132 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
17134 (thunderxt88): Likewise.
17135 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
17136 (thunderx_prefetch_tune): New variable.
17137 (thunderx2t99_prefetch_tune): Update for the correct values.
17138 (thunderxt88_tunings): New variable.
17139 (thunderx_tunings): Use thunderx_prefetch_tune instead of
17140 generic_prefetch_tune.
17141 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
17143 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17145 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
17146 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
17147 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
17148 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
17149 (aarch64_atomic_cas<mode>, GPI): Likewise.
17151 2017-06-21 Martin Liska <mliska@suse.cz>
17153 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
17154 statements on cold and hot labels.
17155 * predict.c (tree_estimate_probability_bb): Remove the
17156 prediction from this place.
17158 2017-06-21 Martin Liska <mliska@suse.cz>
17160 PR tree-optimization/79489
17161 * gimplify.c (maybe_add_early_return_predict_stmt): New
17163 (gimplify_return_expr): Call the function.
17164 * predict.c (tree_estimate_probability_bb): Remove handling
17166 * predict.def: Update comment about early return predictor.
17167 * gimple-predict.h (is_gimple_predict): New function.
17168 * predict.def: Change default value of early return to 66.
17169 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
17171 * passes.def: Put pass_strip_predict_hints to the beginning of
17174 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
17176 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
17177 FUNCTION_DECL declarations.
17178 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
17180 (dwaf2out_decl): Likewise.
17181 * godump.c (go_early_global_decl): Skip call to the real debug hook
17182 for FUNCTION_DECL declarations.
17183 * passes.c (rest_of_decl_compilation): Skip call to the
17184 early_global_decl debug hook for FUNCTION_DECL declarations, unless
17185 -fdump-go-spec is passed.
17187 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
17189 * config/i386/i386.c (struct builtin_isa): New field pure_p.
17190 Reorder for compactness.
17191 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
17192 (def_builtin_pure, def_builtin_pure2): New functions.
17193 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
17195 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
17197 * match.pd (nop_convert): New predicate.
17198 ((A +- CST1) +- CST2): Allow some NOP conversions.
17200 2017-06-21 Jakub Jelinek <jakub@redhat.com>
17203 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
17204 with ctors/dtors if GOVD_SHARED is set.
17206 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17208 * config/aarch64/aarch64.md (movti_aarch64):
17209 Emit mov rather than orr.
17210 (movtf_aarch64): Likewise.
17211 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
17212 Emit mov rather than orr.
17214 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17216 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
17217 Swap alternatives, make integer dup more expensive.
17219 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17221 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
17222 Return true for non-tls symbols.
17224 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
17226 * config/aarch64/aarch64-cores.def (cortex-a55): New.
17227 (cortex-a75): Likewise.
17228 (cortex-a75.cortex-a55): Likewise.
17229 * config/aarch64/aarch64-tune.md: Regenerate.
17230 * doc/invoke.texi (-mtune): Document new values for -mtune.
17232 2017-06-21 Tom de Vries <tom@codesourcery.com>
17234 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
17235 stack_size feature.
17236 (Effective-Target Keywords, Other attributes): Suggest using
17237 dg-add-options stack_size feature to get stack limit in stack_size
17238 effective target documentation.
17240 2017-06-21 Julian Brown <julian@codesourcery.com>
17241 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17243 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
17244 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
17245 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
17247 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
17248 attribute type list for neon_multiply.
17249 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
17250 attribute type list for neon_multiply.
17251 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
17252 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
17253 attribute type list for neon_multiply.
17254 * config/arm/types.md (crypto_pmull): Add.
17255 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
17256 attribute type list.
17258 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
17260 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
17263 2017-06-20 Jakub Jelinek <jakub@redhat.com>
17265 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
17266 to make sure not to dereference a NULL cost_classes_ptr pointer.
17268 2017-06-20 Carl Love <cel@us.ibm.com>
17270 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17271 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17272 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17273 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17274 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17275 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17276 VMULOSW): New enum "unspec" values.
17277 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17278 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17279 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17280 altivec_vmulosw): New patterns.
17281 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17282 VMULOSW): Add definitions.
17284 2017-06-20 Julia Koval <julia.koval@intel.com>
17286 * config/i386/i386.c: Fix rounding expand for new pattern.
17287 * config/i386/subst.md: Fix pattern (parallel -> unspec).
17289 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17291 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
17292 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
17294 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17296 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
17299 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17301 * config/aarch64/aarch64-cores.def: Rearrange to sort by
17302 architecture, then by implementer ID.
17303 * config/aarch64/aarch64-tune.md: Regenerate.
17305 2017-06-20 Richard Biener <rguenther@suse.de>
17307 PR middle-end/81097
17308 * fold-const.c (split_tree): Fold to type before negating.
17310 2017-06-20 David Malcolm <dmalcolm@redhat.com>
17312 * diagnostic-show-locus.c
17313 (selftest::test_fixit_deletion_affecting_newline): New function.
17314 (selftest::diagnostic_show_locus_c_tests): Call it.
17316 2017-06-20 Andreas Schwab <schwab@suse.de>
17319 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
17322 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17324 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
17325 __ARM_FEATURE_COPROC according to support.
17327 2017-06-20 Jakub Jelinek <jakub@redhat.com>
17329 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17330 Rewritten to avoid overflow for > 32-bit pointers.
17333 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
17334 by removing enum keyword.
17335 (ubsan_type_descriptor): Likewise. Formatting fix.
17338 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
17339 splitter): Require TARGET_SSE2 in the condition.
17341 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
17344 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
17345 for doing vector set of SFmode on ISA 3.0.
17346 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
17347 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
17349 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
17350 SFmode value into a V4SF variable that was extracted from another
17351 V4SF variable without converting the element to double precision
17352 and back to single precision vector format.
17353 (vsx_insert_extract_v4sf_p9_2): Likewise.
17355 2017-06-19 Jakub Jelinek <jakub@redhat.com>
17357 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
17358 in UWHI to avoid undefined overflow.
17361 * ubsan.h (enum ubsan_encode_value_phase): New.
17362 (ubsan_encode_value): Change second argument to
17363 enum ubsan_encode_value_phase with default value of
17364 UBSAN_ENCODE_VALUE_GENERIC.
17365 * ubsan.c (ubsan_encode_value): Change second argument to
17366 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
17367 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
17368 create_tmp_var_raw instead of create_tmp_var and use a
17370 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
17371 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
17372 ubsan_encode_value callers.
17375 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
17376 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
17377 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
17379 2017-06-19 Richard Biener <rguenther@suse.de>
17381 PR middle-end/81118
17382 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
17383 estimates if we changed anything.
17385 2017-06-19 Richard Biener <rguenther@suse.de>
17387 PR tree-optimization/80887
17388 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
17389 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
17390 simplified lookups, then reset mprts_hook.
17391 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
17393 (try_to_simplify): Likewise.
17395 2017-06-19 Martin Liska <mliska@suse.cz>
17398 * gimplify.c (gimplify_switch_expr):
17399 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
17401 2017-06-19 Martin Liska <mliska@suse.cz>
17403 * doc/install.texi: Document that PGO runs in 4 stages.
17405 2017-06-19 Martin Liska <mliska@suse.cz>
17408 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
17409 to dispatcher function name.
17410 * multiple_target.c (replace_function_decl): New function.
17411 (create_dispatcher_calls): Redirect both edges and references.
17413 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
17415 * profile-count.c (profile_count::dump): Dump quality.
17416 (profile_count::differs_from_p): Update for unsigned val.
17417 * profile-count.h (profile_count_quality): New enum.
17418 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
17420 2017-06-19 Richard Biener <rguenther@suse.de>
17422 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
17423 struct function as arg.
17424 (estimate_numbers_of_iterations): Export overload with loop arg.
17425 (free_numbers_of_iterations_estimates_loop): Use an overload of
17426 free_numbers_of_iterations_estimates instead.
17427 * tree-cfg.c (remove_bb): Adjust.
17428 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
17429 * tree-parloops.c (gen_parallel_loop): Likewise.
17430 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
17432 (tree_unroll_loops_completely): Likewise.
17433 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
17434 Use an overload instead and export.
17435 (estimated_loop_iterations): Adjust.
17436 (max_loop_iterations): Likewise.
17437 (likely_max_loop_iterations): Likewise.
17438 (estimate_numbers_of_iterations): Take struct function as arg
17440 (loop_exits_before_overflow): Adjust.
17441 (free_numbers_of_iterations_estimates_loop): Use an overload.
17442 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
17443 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
17445 2017-06-19 Richard Biener <rguenther@suse.de>
17448 * ipa-prop.c (find_constructor_constant_at_offset): Handle
17449 RANGE_EXPR conservatively.
17451 2017-06-16 Carl Love <cel@us.ibm.com>
17453 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17454 definitions for vec_float, vec_float2, vec_floato,
17455 vec_floate built-ins.
17456 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
17457 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
17459 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
17460 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
17461 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
17462 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
17463 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
17464 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
17465 vec_floato): Add builtin defines.
17466 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
17467 Update the built-in documentation file for the new built-in
17470 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17472 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
17473 (mthumb): Mark as the negative of -marm.
17475 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17477 * doc/invoke.texi (ARM Options, -mcpu): Document supported
17479 (ARM Options, -mtune): Document that this accepts the same
17480 extension options as -mcpu.
17481 (ARM Options, -mfpu): Document addition of -mfpu=auto.
17483 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17485 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
17486 permitted extensions.
17488 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17490 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
17491 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
17492 (armv8-m.main): Add option +nodsp.
17493 * config/arm/arm-cpu-cdata.h: Regenerated.
17495 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17497 * config/arm/t-fuchsia: New file.
17498 * config.gcc (arm*-*-fuchsia*): Use it.
17500 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17502 * config/arm/t-symbian: Rewrite for new option infrastructure.
17504 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17506 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
17507 (MULTILIB_REQUIRED): Likewise.
17509 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17511 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
17512 (MULTILIB_RESUE): Likewise.
17513 (MULTILIB_MATCHES): Likewise.
17514 (MULTLIB_REQUIRED): Likewise.
17516 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17518 * config/arm/t-rtems: Rewrite for new option framework.
17520 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17522 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
17523 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
17524 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
17525 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
17526 * config/arm/t-multilib: ... here.
17527 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
17528 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
17529 armv7-a and armv8*-a when A-profile libraries have not been built.
17530 * config/arm/t-rmprofile: Rewrite.
17532 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17534 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
17535 with a backslash. Remove the backslash after substituting unescaped
17537 * doc/fragments.texi (MULTILIB_REUSE): Document it.
17539 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17541 * config.gcc: (arm*-*-*): When building a-profile libraries, force
17542 the driver to pass through the default setting of -mfloat-abi.
17543 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
17545 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
17546 (all_feat_combs): New rule.
17547 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
17549 * config/arm/t-aprofile: Rewrite.
17551 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17553 * config/arm/arm.h (FPUTYPE_AUTO): Define.
17554 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
17555 fpu is not specified by the user/command-line.
17556 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
17557 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
17558 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
17559 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
17560 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
17561 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
17563 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17565 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
17566 * config/arm/t-arm-elf: Rewritten.
17568 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17570 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
17571 have some floating-point instructions.
17572 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
17573 (TARGET_MAYBE_HARD_FLOAT): New macro.
17574 * config/arm/arm-builtins.c (arm_init_builtins): Use
17575 TARGET_MAYBE_HARD_FLOAT.
17576 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
17578 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17580 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
17581 (configargs.h): Include it.
17582 (arm_print_hint_for_fpu_option): New function.
17583 (arm_parse_fpu_option): New function.
17584 (candidate_extension): New class.
17585 (arm_canon_for_multilib): New function.
17586 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
17587 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
17588 (ARCH_CANONICAL_SPECS): New macro.
17589 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
17591 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17593 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
17594 are set after handling multilib fragments. Set target_cpu_default2
17597 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17599 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
17601 (arm*-*-*): Set target_cpu_default2 to a quoted string.
17602 * config/arm/parsecpu.awk (check_cpu): Validate any extension
17604 (check_arch): Likewise.
17605 * config/arm/arm.c (arm_configure_build_target): Handle
17606 TARGET_CPU_DEFAULT being a string constant. Scan any feature
17607 options in the default.
17609 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17611 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
17612 when an option is an alias of another.
17613 * config/arm/parsecpu.awk (optalias): New parser token.
17614 (gen_comm_data): Mark non-alias options as such. Emit entries
17615 for extension aliases.
17616 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
17617 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
17618 (armv6kz, armv6zk, armv6t2): Likewise.
17619 (armv7): Make vfpv3-d16 an alias.
17620 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
17622 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
17623 Sort in canonical order.
17624 (armv8-a): Sort in canonical order.
17625 (armv8.1-a, armv8.2-a): Likewise.
17626 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
17628 (cortex-a9): Sort in canonical order.
17629 * config/arm/arm.c (selftests.h): Include it.
17630 (arm_test_cpu_arch_data): New function.
17631 (arm_run_self_tests): New function.
17632 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
17633 (targetm): Move declaration to the end of the file.
17634 * arm-cpu-cdata.h: Regenerated.
17636 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17638 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
17639 call to target_mode_check describing the type of option passed.
17640 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
17641 (arm_target_thumb_only): Use arm_parse_arch_option_name or
17642 arm_parse_cpu_option_name to match parameters against list of
17644 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
17645 arm_arch_core_flags data structure.
17646 * config/arm/arm-cpu_cdata.h: Regenerated.
17648 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17650 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
17652 (arm_print_hint_for_cpu_option): Likewise.
17653 (arm_print_hint_for_arch_option): Likewise.
17654 (arm_parse_cpu_option_name): Likewise.
17655 (arm_parse_arch_option_name): Likewise.
17656 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
17657 of entries in the all_fpus list.
17658 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
17659 (arm_parse_cpu_option_name): Declare.
17660 (arm_parse_arch_option_name): Declare.
17661 (arm_parse_option_features): Declare.
17662 (arm_intialize_isa): Declare.
17663 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
17665 (gen_comm_data): ... here. Make definitions non-static.
17666 * config/arm/arm-cpu-data.h: Regenerated.
17667 * config/arm/arm-cpu-cdata.h: Regenerated.
17669 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17671 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
17672 (cpu_arch_extension): New structure.
17673 (cpu_arch_option, arch_option, cpu_option): New structures.
17674 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
17675 architecture types.
17676 (gen_data): Generate new format data tables.
17677 * config/arm/arm.c (cpu_tune): New structure.
17678 (cpu_option, processors): Delete.
17679 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
17680 (arm_print_hint_for_cpu_option): ... this and ...
17681 (arm_print_hint_for_arch_option): ... this.
17682 (arm_parse_arch_cpu_name): Delete. Replace with ...
17683 (arm_parse_cpu_option_name): ... this and ...
17684 (arm_parse_arch_option_name): ... this.
17685 (arm_unrecognized_feature): Change type of target parameter to
17687 (arm_parse_arch_cpu_features): Delete. Replace with ...
17688 (arm_parse_option_features): ... this.
17689 (arm_configure_build_target): Rework to use new configuration data
17691 (arm_print_tune_info): Rework for new configuration data tables.
17692 * config/arm/arm-cpu-data.h: Regenerated.
17693 * config/arm/arm-cpu.h: Regenerated.
17695 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17697 * Makefile.in (OBJS): Move sbitmap.o from here ...
17698 (OBJS-libcommon): ... to here.
17700 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17702 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
17703 (ISA_ALL_CRYPTO): New macro.
17704 (ISA_ALL_SIMD): New macro
17705 (ISA_ALL_FP): New macro.
17706 * config/arm/arm.c (fpu_bitlist): Update initializer.
17707 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
17709 (arm9e): Add fpu. Add option for nofp
17710 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
17711 (arm926ej-s, arm1026ej-s): Likewise.
17712 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
17713 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
17714 neon-fp16, neon-vfpv4, nofp and nosimd.
17715 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
17716 (cortex-a8): Add fpu. Add option for nofp.
17717 (cortex-a9): Add fpu. Add options for nosimd and nofp.
17718 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
17719 (cortex-r4f): Add fpu.
17720 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
17721 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
17723 (cortex-r8): Likewise.
17724 (cortex-m4): Add fpu. Add option for nofp.
17725 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
17726 (cortex-a17.cortex-a7): Likewise.
17727 (cortex-a32): Add fpu. Add options for crypto and nofp.
17728 (cortex-a35, cortex-a53): Likewise.
17729 (cortex-a57): Add fpu. Add option for crypto.
17730 (cortex-a72, cortex-a73): Likewise.
17731 (exynos-m1): Likewise.
17732 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
17733 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
17734 (cortex-m33): Add fpu. Add option for nofp.
17735 * config/arm/arm-cpu-cdata.h: Regenerated
17736 * config/arm/arm-cpu-data.h: Regenerated.
17738 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17740 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
17741 (armv5te, armv5tej): Likewise.
17742 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
17743 (armv7): Add options fp and vfpv3-d16.
17744 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
17745 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
17747 (armv7ve): Likewise.
17748 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
17749 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
17750 (armv8-a): Add nocrypto option.
17751 (armv8.1-a, armv8.2-a): Likewise.
17752 (armv8-m.main): add options fp, fp.dp and nofp.
17754 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17756 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
17758 (armv8-a+crc): Delete.
17759 (armv8.1-a): Add options simd, crypto and nofp.
17760 (armv8.2-a): Add options fp16, simd, crypto and nofp.
17761 (armv8.2-a+fp16): Delete.
17762 (armv8-m.main): Add option dsp.
17763 (armv8-m.main+dsp): Delete.
17764 (cortex-a8): Add fpu. Add nofp option.
17765 (cortex-a9): Add fpu. Add nofp and nosimd options.
17766 * config/arm/parsecpu.awk (gen_data): Generate option tables and
17767 link to main cpu and architecture data structures.
17768 (gen_comm_data): Only put isa attributes from the main architecture
17770 (option): New statement for architecture and CPU entries.
17771 * arm.c (struct cpu_option): New structure.
17772 (struct processors): Add entry for options.
17773 (arm_unrecognized_feature): New function.
17774 (arm_parse_arch_cpu_name): Ignore any characters after the first
17776 (arm_parse_arch_cpu_feature): New function.
17777 (arm_configure_build_target): Separate out any CPU and architecture
17778 features and parse separately. Don't error out if -mfpu=auto is
17779 used with only an architecture string.
17780 (arm_print_asm_arch_directives): New function.
17781 (arm_file_start): Call it.
17782 * config/arm/arm-cpu-cdata.h: Regenerated.
17783 * config/arm/arm-cpu-data.h: Likewise.
17784 * config/arm/arm-tables.opt: Likewise.
17786 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17788 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
17789 assembler when it is not -mfpu=auto.
17791 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17793 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
17794 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
17795 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
17796 (ASM_CPU_SPEC): Rewrite.
17797 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
17798 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
17799 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
17800 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
17801 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
17802 copied string is NUL-terminated. Also strip any characters prefixed
17804 (arm_rewrite_selected_arch): New function.
17805 (arm_rewrite_march): New function.
17807 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17809 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
17810 (x_arm_cpu_string, x_arm_tune_string): Likewise.
17811 (march, mcpu, mtune): Convert to string-based options.
17812 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
17813 (arm_parse_arch_cpu_name): New function.
17814 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
17815 identify selected architecture or CPU.
17816 (arm_option_save): New function.
17817 (TARGET_OPTION_SAVE): Redefine.
17818 (arm_option_restore): Restore string options.
17819 (arm_option_print): Print string options.
17821 2017-06-16 Martin Sebor <msebor@redhat.com>
17823 PR tree-optimization/80933
17824 PR tree-optimization/80934
17825 * builtins.c (fold_builtin_3): Do not handle bcmp here.
17826 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
17827 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
17828 (gimple_fold_builtin): Call them.
17830 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17832 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
17833 as unlikely; update profile.
17835 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17837 * predict.c (force_edge_cold): Handle declaring edges impossible
17840 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17842 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
17844 (try_unroll_loop_completely): Fix reporting.
17846 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17848 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
17850 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
17853 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
17854 if given a non-constant argument for an intrinsic which requires a
17857 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17859 * profile.c (compare_freqs): New function.
17860 (branch_prob): Sort edge list.
17861 (find_spanning_tree): Assume that the list is priority sorted.
17863 2017-06-16 Richard Biener <rguenther@suse.de>
17865 PR tree-optimization/81090
17866 * passes.def (pass_record_bounds): Remove.
17867 * tree-pass.h (make_pass_record_bounds): Likewise.
17868 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
17869 make_pass_record_bounds): Likewise.
17870 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
17871 not free niter estimates at the beginning but at the end.
17872 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
17874 2017-06-16 Richard Biener <rguenther@suse.de>
17876 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
17877 initializer to workaround ICE in host GCC 4.8.
17879 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17881 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
17883 (clone_inlined_nodes): Update.
17885 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17887 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
17888 prefetch settings, and enable prefetching by default at -O3.
17890 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17892 * config/aarch64/aarch64.c (aarch64_override_options_internal):
17893 Set flag_prefetch_loop_arrays according to tuning data.
17895 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17897 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
17898 New tune structure.
17899 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
17900 [Unrelated to main purpose of the patch] Place the pointer field last
17901 to enable type checking errors when tune structure are wrongly merged.
17902 * config/aarch64/aarch64.c (generic_prefetch_tune,)
17903 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
17904 (thunderx2t99_prefetch_tune): New tune constants.
17905 (tune_params *_tunings): Update all tunings (no functional change).
17906 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
17907 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
17908 from tunings structures.
17910 2017-06-16 Jakub Jelinek <jakub@redhat.com>
17913 * ubsan.c (instrument_null): Add T argument, use it instead
17914 of computing it based on IS_LHS.
17915 (instrument_object_size): Likewise.
17916 (pass_ubsan::execute): Adjust instrument_null and
17917 instrument_object_size callers to pass gimple_get_lhs or
17918 gimple_assign_rhs1 result to it. Use instrument_null instead of
17919 calling get_base_address and instrument_mem_ref. Handle
17920 aggregate call arguments for object-size sanitization.
17922 2017-06-16 Yury Gribov <tetra2005@gmail.com>
17924 PR tree-optimization/81089
17925 * tree-vrp.c (is_masked_range_test): Validate operands of
17928 2017-06-15 Martin Sebor <msebor@redhat.com>
17931 * dumpfile.c (dump_register): Avoid calling memset to initialize
17932 a class with a default ctor.
17933 * gcc.c (struct compiler): Remove const qualification.
17934 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
17935 * hash-table.h: Ditto.
17936 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
17938 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
17939 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
17941 * params.h (struct param_info): Make struct members non-const.
17942 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
17943 with default initialization.
17944 * vec.h (vec_copy_construct, vec_default_construct): New helper
17946 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
17947 with vec_copy_construct.
17948 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
17949 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
17950 * doc/invoke.texi (-Wclass-memaccess): Document.
17952 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17954 * emit-rtl.h (is_leaf): Update comment about local
17955 register allocator.
17957 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17960 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
17961 for a variable to have a section before checking if the section has a
17963 Set section to.persistent if persistent attribute is set.
17964 Warn if .persistent attribute is used on an automatic variable.
17966 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
17968 PR rtl-optimization/80474
17969 * reorg.c (update_block): Do not ignore instructions in a delay slot.
17971 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
17973 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
17976 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
17978 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
17979 (casesi): Emit bounds checking as RTL.
17980 (casesi_internal_mips16_<mode>): Remove bounds checking.
17982 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
17984 * config/xtensa/xtensa.c (xtensa_option_override): Append
17985 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
17986 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
17987 xtensa_doloop_hooks): Define unconditionally.
17988 (xtensa_reorg_loops): Only call reorg_loops in the presence of
17990 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
17991 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
17992 for it in xtensa_option_override.
17993 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
17994 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
17996 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
17998 * doc/cppopts.texi: Document '-' special value to -MF.
18000 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
18002 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
18003 (cortex_a53_fconst): Likewise.
18004 (cortex_a53_fpmul): Likewise.
18005 (cortex_a53_f_load_64): Likewise.
18006 (cortex_a53_f_load_many): Likewise.
18007 (cortex_a53_advsimd_alu): Likewise.
18008 (cortex_a53_advsimd_alu_q): Likewise.
18009 (cortex_a53_advsimd_mul): Likewise.
18010 (cortex_a53_advsimd_mul_q): Likewise.
18011 (fpmac bypass): Add new bypass for fpmac-fpmac case.
18012 Add missing fmul, r2f_cvt and fconst cases.
18014 2017-06-14 Richard Biener <rguenther@suse.de>
18016 PR middle-end/81088
18017 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
18019 (fold_binary_loc): When associating do not treat pre-existing
18020 TREE_OVERFLOW on literal constants as a reason to allow
18021 TREE_OVERFLOW on associated literal constants.
18023 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18025 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
18026 (MASK_FEATURES): New macro.
18027 * config/sparc/sparc.c (sparc_option_override): Remove the special
18028 handling of -mfpu and generalize it to all MASK_FEATURES switches.
18030 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18032 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
18033 a division of 0 if non-call exceptions are enabled.
18035 2017-06-14 Andrew Pinski <apinski@cavium.com>
18036 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18039 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
18040 Improve vector initialization code gen for only variable case.
18042 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18044 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
18046 2017-06-14 Richard Biener <rguenther@suse.de>
18048 PR tree-optimization/81083
18049 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
18052 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18054 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
18055 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
18056 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
18057 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
18058 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
18059 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
18061 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18063 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
18064 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
18066 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18068 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
18070 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18072 * config/rs6000/t-rtems: Don't handle SPE.
18074 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18076 * config/rs6000/t-linux: Don't handle SPE.
18078 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18080 * config/rs6000/eabispe.h: Delete file.
18082 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18084 * config/rs6000/t-spe: Delete file.
18086 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18088 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
18089 (rs6000_legitimate_offset_address_p): Return false for anything in
18090 V2SImode or V2SFmode.
18092 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18094 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
18095 except V2SF and V2SI. Rearrange the vector modes, and add comments.
18096 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
18098 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
18099 (rs6000_legitimate_offset_address_p): Ditto.
18100 (rs6000_emit_move): Ditto.
18101 (rs6000_init_builtins): Remove V4HI_type_node.
18103 2017-06-13 Martin Liska <mliska@suse.cz>
18106 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
18107 (gate_asan): Likewise.
18108 * asan.h (asan_no_sanitize_address_p): Remove the function.
18109 (sanitize_flags_p): New function.
18110 * builtins.def: Fix coding style.
18111 * common.opt: Use renamed enum value.
18112 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
18113 * doc/extend.texi: Document no_sanitize attribute.
18114 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
18115 to SANITIZE_UNDEFINED_NONDEFAULT.
18116 * gcc.c (sanitize_spec_function): Use the renamed enum value.
18117 * gimple-fold.c (optimize_atomic_compare_exchange_p):
18118 Use sanitize_flags_p.
18119 * gimplify.c (gimplify_function_tree): Likewise.
18120 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
18121 * opts.c (parse_no_sanitize_attribute): New function.
18122 (common_handle_option): Use renamed enum value.
18123 * opts.h (parse_no_sanitize_attribute): Declare.
18124 * tree.c (sanitize_flags_p): New function.
18125 * tree.h: Declared here.
18126 * tsan.c: Use sanitize_flags_p.
18127 * ubsan.c (ubsan_expand_null_ifn): Likewise.
18128 (instrument_mem_ref): Likewise.
18129 (instrument_bool_enum_load): Likewise.
18130 (do_ubsan_in_current_function): Remove the function.
18131 (pass_ubsan::execute): Use sanitize_flags_p.
18132 * ubsan.h: Remove do_ubsan_in_current_function
18133 * tree-cfg.c (print_no_sanitize_attr_value): New function.
18134 (dump_function_to_file): Use it here.
18136 2017-06-13 Martin Jambor <mjambor@suse.cz>
18138 PR tree-optimization/80803
18139 PR tree-optimization/81063
18140 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
18141 (propagate_subaccesses_across_link): Enqueue subtree whenever
18142 necessary instead of relying on the caller.
18144 2017-06-13 Martin Jambor <mjambor@suse.cz>
18146 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
18147 that have a first_link.
18148 (sort_and_splice_var_accesses): Do not check first_link before
18150 (subtree_mark_written_and_enqueue): Likewise.
18151 (propagate_all_subaccesses): Likewise and do not stop at first
18152 parent with a first_link.
18154 2017-06-13 Martin Jambor <mjambor@suse.cz>
18156 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
18159 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18161 * match.pd: New pattern.
18163 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18165 * tree-vrp.c (is_masked_range_test): New function.
18166 (register_edge_assert_for): Determine ranges for
18169 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18171 PR tree-optimization/67328
18172 * fold-const.c (maskable_range_p): New function.
18173 (build_range_check): Generate bittests if possible.
18175 2017-06-13 Martin Liska <mliska@suse.cz>
18177 * gimple-pretty-print.c (dump_probability): Add new argument.
18178 (dump_edge_probability): Dump both probability and count.
18179 (dump_gimple_label): Likewise.
18180 (dump_gimple_bb_header): Likewise.
18182 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
18185 * config/avr/avr-devices.c: Fix indentation.
18186 * config/avr/gen-avr-mmcu-specs.c: Dito.
18188 2017-06-13 Richard Biener <rguenther@suse.de>
18190 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
18191 instead get vector type from stmt_info.
18192 (vectorizable_reduction): Adjust. Remove dead code.
18194 2017-06-13 Richard Biener <rguenther@suse.de>
18196 PR middle-end/81065
18197 * fold-const.c (extract_muldiv_1): Remove bogus distribution
18198 case of C * (x * C2 + C3).
18199 (fold_addr_of_array_ref_difference): Properly fold index difference.
18201 2017-06-12 David S. Miller <davem@davemloft.net>
18204 * config/sparc/sparc.md (return expander): Emit frame blockage if
18205 function uses alloca.
18207 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
18209 * combine.c (make_field_assignment): Check len rather than the mode
18210 precision when calling force_to_mode.
18212 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
18214 Support multilibs and devices that see flash in RAM address range.
18217 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
18218 (avr_mcu_t) <flash_pm_offset>: New field.
18219 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
18220 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
18221 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
18222 (AVR_TINY_PM_OFFSET): Remove macro.
18223 * config/avr/avr.opt (-mshort-calls): New option.
18224 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18225 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
18226 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
18227 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
18228 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
18229 instead of avr_arch->have_jmp_call.
18230 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
18231 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
18232 avr_arch->flash_pm_offset to define.
18233 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
18234 new field flash_pm_offset. Add entry for avrxmega3.
18235 (avr_texinfo): Add entry for avrxmega3.
18236 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
18237 attiny212, attiny214,
18238 attiny412, attiny414, attiny416, attiny417,
18239 attiny814, attiny816, attiny817,
18240 attiny1614, attiny1616, attiny1617,
18241 attiny3214, attiny3216, attiny3217.
18242 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
18243 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
18244 (avr_print_operand_address) [AVR_TINY]: Same.
18245 (avr_asm_init_sections) <readonly_data_section>: Only patch
18246 callback if avr_arch->flash_pm_offset = 0.
18247 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
18248 for rodata if avr_arch->flash_pm_offset != 0.
18249 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
18250 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
18251 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
18252 (m_options): Append opt_rcall.
18253 (m_dirnames): Append dir_rcall.
18254 * config/avr/t-multilib: Regenerate.
18256 * configure.ac [target=avr]: Check whether avrxmega3 default
18257 linker description file works as needed.
18258 * configure: Regenerate.
18259 * doc/avr-mmcu.texi: Regenerate.
18260 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
18261 <__AVR_ARCH__>: Document avrxmega3 and 103.
18262 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
18263 <__AVR_SHORT_CALLS__>: Document it.
18264 <__AVR_PM_BASE_ADDRESS__>: Document it.
18265 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
18266 (AVR Variable Attributes) <progmem>: Document this is
18267 not needed for avrxmega3.
18268 (AVR Named Address Spaces) <__flash>: Dito.
18270 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
18272 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
18274 2017-06-12 Doug Rupp <rupp@adacore.com>
18276 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
18277 Append vxworks-stdint.h to the tm_file list.
18278 * config/vxworks-stdint.h: New file.
18280 2017-06-12 Martin Liska <mliska@suse.cz>
18282 PR tree-optimization/81041
18283 * tree-profile.c (gimple_gen_ic_func_profiler):
18284 Create an extra BB in profile-generate
18285 (gimple_gen_time_profiler): Likewise.
18287 2017-06-12 Jakub Jelinek <jakub@redhat.com>
18289 PR tree-optimization/81003
18290 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
18291 (update_range_test): Use it instead of force_gimple_operand_gsi.
18293 2017-06-12 Richard Biener <rguenther@suse.de>
18295 PR tree-optimization/81053
18296 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
18297 with backedge value not defined in loop. Simplify def stmt
18300 2017-06-11 Tom de Vries <tom@codesourcery.com>
18303 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
18305 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
18306 nvptx_cannot_force_const_mem.
18308 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18310 * opts.c (finish_options): Move test for flag_split_stack after
18311 it has been initialized.
18313 2017-06-11 Jason Merrill <jason@redhat.com>
18315 * tree.h (id_equal): New.
18316 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
18317 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
18318 instead of strcmp of IDENTIFIER_POINTER.
18320 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18322 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
18323 (mark_all_inlined_calls_cdtor): Fix formating.
18324 (inline_transform): Rescale profile before inlining.
18326 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18328 * cgraph.h (cgraph_edge::clone): Update prototype.
18329 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
18330 (cgraph_node::create_clone): Update.
18331 (cgraph_node::create_version_clone): Update.
18332 * tree-inline.c (copy_bb): Update.
18333 (expand_call_inline): Update.
18335 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
18337 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
18338 factored out from ...
18339 (rs6000_emit_prologue): ... here.
18341 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
18343 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
18344 factored out from ...
18345 (rs6000_emit_prologue): ... here.
18347 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18349 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
18351 (handle_missing_profiles): Fix computation of tp_first_run.
18352 (counts_to_freqs): Do not touch freqs when count is 0.
18354 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18356 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
18359 2017-06-10 Tom de Vries <tom@codesourcery.com>
18361 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18362 attributes): Document signal effective target.
18364 2017-06-10 Tom de Vries <tom@codesourcery.com>
18366 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18367 Document effective target stack_size.
18369 2017-06-09 David Malcolm <dmalcolm@redhat.com>
18371 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
18372 to the edit_context if they can be auto-applied.
18374 2017-06-9 Ian Lance Taylor <iant@golang.org>
18376 * opts.c (finish_options): If -fsplit-stack, disable implicit
18377 -forder-blocks-and-partition.
18378 * doc/invoke.texi (Optimize Options): Document that when using
18379 -fsplit-stack -forder-blocks-and-partition is not implicitly
18382 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18384 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
18385 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
18386 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
18387 * builtins.def (abort, trap, unreachable): Declare cold.
18388 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
18389 * tree-core.h (ECF_COLD): New.
18390 * tree.c (set_call_expr_flags): Handle ECF_COLD.
18391 (build_common_builtin_nodes): Mark unreachable and abort as cold.
18393 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18395 * predict.c (unlikely_executed_stmt_p): Cleanup.
18397 2017-06-09 Richard Biener <rguenther@suse.de>
18399 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
18400 model if the ref is always written to.
18402 2017-06-09 Tamar Christina <tamar.christina@arm.com>
18404 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
18406 2017-06-09 Tamar Christina <tamar.christina@arm.com>
18408 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
18411 2017-06-09 Tom de Vries <tom@codesourcery.com>
18414 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
18415 "target cannot support label values" when encountering LABEL_REF.
18417 2017-06-09 Martin Liska <mliska@suse.cz>
18419 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
18420 (gimple_gen_ic_func_profiler): Emit direct comparison
18421 of __gcov_indirect_call_callee with NULL.
18422 (gimple_gen_time_profiler): Change probability from
18423 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
18425 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18427 * profile.c (edge_gcov_counts): Turn to pointer.
18428 (compute_branch_probabilities, compute_branch_probabilities): Update.
18429 (branch_prob): Do not clear edge_gcov_count.
18430 * profile.h (edge_gcov_counts): Turn to pointer.
18431 (edge_gcov_count): Update.
18433 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18435 * gimple.h (gimple_check_failed): Mark cold.
18437 2017-06-09 Richard Biener <rguenther@suse.de>
18439 PR tree-optimization/66623
18440 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
18441 refactor check_reduction into two parts, properly computing
18442 whether we have to check reduction validity for outer loop
18445 2017-06-09 Richard Biener <rguenther@suse.de>
18447 PR tree-optimization/79483
18448 * graphite-scop-detection.c (order): New global.
18449 (get_order): Compute bb to order mapping that satisfies code
18450 generation constraints.
18451 (cmp_pbbs): New helper.
18452 (build_scops): Start domwalk at entry block, sort generated
18455 2017-06-09 Richard Biener <rguenther@suse.de>
18457 PR middle-end/81007
18458 * ipa-polymorphic-call.c
18459 (ipa_polymorphic_call_context::restrict_to_inner_class):
18460 Skip FIELD_DECLs with error_mark_node type.
18461 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
18464 2017-06-09 Martin Liska <mliska@suse.cz>
18466 * predict.c (struct branch_predictor): New struct.
18467 (test_prediction_value_range): New test.
18468 (predict_c_tests): New function.
18469 * selftest-run-tests.c (selftest::run_tests): Run the function.
18470 * selftest.h: Declare new tests.
18472 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
18475 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
18476 gen_add3_insn did not fail.
18477 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
18478 r0, construct that number in a temporary reg and add that reg to r0.
18479 If asked to put the result in r0 as well, fail.
18481 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
18483 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18484 for early expansion of vec_eqv.
18486 2017-06-08 Jakub Jelinek <jakub@redhat.com>
18488 PR middle-end/81005
18489 * ubsan.c (instrument_null): Avoid pointless code temporary.
18490 (pass_ubsan::execute): Instrument aggregate arguments of calls.
18492 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
18496 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
18499 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
18502 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18504 * predict.c (unlikely_executed_edge_p): Move ahead.
18505 (probably_never_executed_edge_p): Use it.
18507 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18509 PR middle-end/79988
18510 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
18511 gimple_call_builtin_p call.
18513 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18515 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
18516 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
18517 rtl_check_failed_type2, rtl_check_failed_code1,
18518 rtl_check_failed_code2, rtl_check_failed_code_mode,
18519 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
18520 rtvec_check_failed_bounds, rtl_check_failed_flag,
18521 _fatal_insn_not_found, _fatal_insn): Likewise.
18522 * tree.h (tree_contains_struct_check_failed,
18523 tree_check_failed, tree_not_check_failed,
18524 tree_class_check_failed, tree_range_check_failed,
18525 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
18526 tree_vec_elt_check_failed, phi_node_elt_check_failed,
18527 tree_operand_check_failed, omp_clause_check_failed,
18528 omp_clause_operand_check_failed, omp_clause_range_check_failed):
18531 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18533 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
18534 flag_branch_probabilities.
18535 * ipa-inline.c (edge_badness): Likewise.
18536 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
18537 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
18538 * predict.c (maybe_hot_frequency_p): Likewise.
18539 (probably_never_executed): Likewise.
18540 * sched-ebb.c (schedule_ebbs): Likewise.
18541 * sched-rgn.c (find_single_block_region): Likewise.
18542 * tracer.c (tail_duplicate): Likewise.
18544 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18546 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
18547 longer requires x_flag_profile_use.
18549 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18551 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
18552 instead of flag_reorder_blocks_and_partition.
18553 * dbxout.c (dbxout_function_end): Likewise.
18554 * dwarf2out.c (gen_subprogram_die): Likewise.
18555 * haifa-sched.c (sched_create_recovery_edges): Likewise.
18556 * hw-doloop.c (reorg_loops): Likewise.
18557 * varasm.c (assemble_start_function,
18558 assemble_end_function): Likewise.
18559 (decide_function_section): Do not check for
18560 flag_reorder_blocks_and_partition.
18562 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18564 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18566 (chkp_get_hard_register_fake_addr_expr): Ditto.
18567 (chkp_build_addr_expr): Add check for hard reg case.
18568 (chkp_parse_array_and_component_ref): Ditto.
18569 (chkp_find_bounds_1): Ditto.
18570 (chkp_process_stmt): Don't generate bounds store for
18573 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18575 * predict.c (maybe_hot_bb_p): Do not check profile status.
18576 (maybe_hot_edge_p): Likewise.
18577 (probably_never_executed): Check for zero counts even if profile
18579 (unlikely_executed_edge_p): New function.
18580 (unlikely_executed_stmt_p): New function.
18581 (unlikely_executed_bb_p): New function.
18582 (set_even_probabilities): Use unlikely predicates.
18583 (combine_predictions_for_bb): Likewise.
18584 (predict_paths_for_bb): Likewise.
18585 (predict_paths_leading_to_edge): Likewise.
18586 (determine_unlikely_bbs): New function.
18587 (estimate_bb_frequencies): Use it.
18588 (compute_function_frequency): Use zero counts even if profile is
18590 * profile-count.h: Fix typo.
18592 2017-08-08 Julia Koval <julia.koval@intel.com>
18594 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
18595 _mm512_mask_cvtsepi16_storeu_epi8,
18596 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
18597 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
18598 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
18599 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
18600 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
18601 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
18602 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
18603 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
18604 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
18605 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
18606 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
18607 __builtin_ia32_pmovuswb256mem_mask,
18608 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
18609 __builtin_ia32_pmovwb512mem_mask): New builtins.
18611 2017-08-08 Julia Koval <julia.koval@intel.com>
18613 PR target/73350,80862
18614 * config/i386/subst.md (round): Fix round pattern.
18615 * config/i386/i386.c (ix86_erase_embedded_rounding):
18616 Fix erasing rounding for the fixed pattern.
18618 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18620 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
18622 2017-06-08 Martin Liska <mliska@suse.cz>
18624 PR gcov-profile/80911
18625 * gcov.c (block_info::block_info): New constructor.
18627 2017-06-07 Carl Love <cel@us.ibm.com>
18629 * config/rs6000/rs6000-c: The return type of the following
18630 built-in functions was implemented as int not long long. Fix sign
18631 of return value for the unsigned version of vec_mulo and vec_mule.
18632 vector unsigned long long vec_bperm (vector unsigned long long,
18633 vector unsigned char)
18634 vector signed long long vec_mule (vector signed int,
18636 vector unsigned long long vec_mule (vector unsigned int,
18637 vector unsigned int)
18638 vector signed long long vec_mulo (vector signed int,
18640 vector unsigned long long vec_mulo (vector unsigned int,
18641 vector unsigned int)
18642 * doc/extend.texi: Fix the documentation for the built-in
18645 2017-06-07 Carl Love <cel@us.ibm.com>
18648 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
18651 2017-06-07 Carl Love <cel@us.ibm.com>
18653 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
18654 support, Generate doublehv for signed int/float for BE case only.
18656 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
18658 * doc/invoke.texi (mcx16): Rewrite.
18660 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18662 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
18663 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
18664 *mov<mode>_softfloat, and an anonymous splitter): Use
18665 nonimmediate_operand instead of rs6000_nonimmediate_operand.
18667 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18669 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
18671 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18672 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
18673 (rs6000_debug_reg_global): Adjust.
18674 (rs6000_init_hard_regno_mode_ok): Adjust.
18675 (rs6000_dbx_register_number): Adjust.
18676 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
18677 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18678 Remove SPE_ACC and SPEFSCR.
18679 (REG_ALLOC_ORDER): Ditto.
18680 (FRAME_POINTER_REGNUM): Change to 111.
18681 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
18682 (REG_CLASS_NAMES): Ditto.
18683 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
18684 (REGISTER_NAMES): Ditto.
18685 (ADDITIONAL_REG_NAMES): Ditto.
18686 (rs6000_reg_names): Ditto.
18687 * config/rs6000/rs6000.md: Renumber some register number
18690 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18692 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
18694 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18695 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
18697 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
18698 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18699 Delete the SPE high registers.
18700 (REG_ALLOC_ORDER): Ditto.
18701 (enum reg_class): Remove SPE_HIGH_REGS.
18702 (REG_CLASS_NAMES): Ditto.
18703 (REG_CLASS_CONTENTS): Delete the SPE high registers.
18704 (REGISTER_NAMES): Ditto.
18705 (rs6000_reg_names): Ditto.
18706 * doc/tm.texi.in: Remove SPE as example.
18707 * doc/tm.texi: Regenerate.
18709 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18711 * config/rs6000/8540.md (ppc8540_brinc): Delete.
18712 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
18713 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
18714 * config/rs6000/rs6000.md (type): Remove "brinc".
18716 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18718 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
18719 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
18720 * config/rs6000/linuxspe.h: Delete file.
18721 * config/rs6000/rs6000.md: Don't include spe.md.
18722 * config/rs6000/spe.h: Delete file.
18723 * config/rs6000/spe.md: Delete file.
18724 * config/rs6000/t-rs6000: Remove spe.md.
18726 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18728 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
18729 (reg_or_none500mem_operand): Delete.
18730 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
18731 instead of reg_or_none500mem_operand.
18733 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18735 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18736 handling of SPE flags.
18737 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
18739 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18741 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
18743 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
18744 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
18745 paired_divv2sf3): Similar.
18746 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
18747 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
18748 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
18750 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
18751 Rename the paired_* instruction patterns.
18752 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
18754 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
18755 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
18756 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
18757 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
18758 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
18759 PAIRED_VECTOR_MODE.
18760 (struct machine_function): Delete field spe_insn_chain_scanned_p.
18761 (spe_func_has_64bit_regs_p): Delete.
18762 (spe_expand_predicate_builtin): Delete.
18763 (spe_expand_evsel_builtin): Delete.
18764 (TARGET_DWARF_REGISTER_SPAN): Do not define.
18765 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
18766 (invalid_e500_subreg): Delete.
18767 (rs6000_legitimize_address): Always force_reg op2 as well, for
18768 paired single memory accesses.
18769 (rs6000_member_type_forces_blk): Delete.
18770 (rs6000_spe_function_arg): Delete.
18771 (rs6000_expand_unop_builtin): Delete SPE handling.
18772 (rs6000_expand_binop_builtin): Ditto.
18773 (spe_expand_stv_builtin): Delete.
18774 (bdesc_2arg_spe): Delete.
18775 (spe_expand_builtin): Delete.
18776 (spe_expand_predicate_builtin): Delete.
18777 (spe_expand_evsel_builtin): Delete.
18778 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
18779 (spe_init_builtins): Delete.
18780 (spe_func_has_64bit_regs_p): Delete.
18781 (savres_routine_name): Delete "info" parameter. Adjust callers.
18782 (rs6000_emit_stack_reset): Ditto.
18783 (rs6000_dwarf_register_span): Delete.
18784 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
18785 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
18786 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
18788 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
18790 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
18791 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
18792 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
18793 mulv2sf3, divv2sf3): Delete expanders.
18795 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18797 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
18799 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18801 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
18802 * config/rs6000/rs6000.c: Ditto.
18804 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18806 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
18807 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
18808 comparison_operator.
18810 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18812 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
18813 * config/rs6000/rs6000.opt: Ditto.
18814 * config/rs6000/t-rtems: Ditto.
18816 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18818 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
18819 TARGET_E500_SINGLE by 0, simplify.
18820 * config/rs6000/rs6000.c: Ditto.
18821 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
18822 (spe_build_register_parallel): Delete.
18823 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
18824 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
18825 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
18826 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
18827 (E500_CONVERT): Delete.
18828 * config/rs6000/spe.md: Remove many patterns and all define_constants.
18830 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18832 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
18833 * config/rs6000/dfp.md: Ditto.
18834 (negdd2, *negdd2_fpr): Merge.
18835 (absdd2, *absdd2_fpr): Merge.
18836 (negtd2, *negtd2_fpr): Merge.
18837 (abstd2, *abstd2_fpr): Merge.
18838 * config/rs6000/e500.h: Delete file.
18839 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
18840 TARGET_FPRS by 1 and simplify.
18841 * config/rs6000/rs6000-c.c: Ditto.
18842 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
18843 TARGET_DF_SPE by 0.
18844 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
18846 * config/rs6000/rs6000.md: Ditto.
18847 (floatdidf2, *floatdidf2_fpr): Merge.
18848 (move_from_CR_gt_bit): Delete.
18849 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
18850 (E500_CR_IOR_COMPARE): Delete.
18851 (All patterns that require !TARGET_FPRS): Delete.
18852 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
18854 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18856 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
18858 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18860 * graphds.c (add_edge): Intitialize edge's attached data.
18861 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
18862 pointer parameter. Call pointed function on each edge during
18863 graph traversing. Skip traversing the edge when the function
18865 (graphds_dfs, graphds_scc): Ditto.
18866 (for_each_edge): New parameter. Pass the new parameter to callback
18868 * graphds.h (skip_edge_callback): New function pointer type.
18869 (graphds_dfs, graphds_scc): New function pointer parameter.
18870 (graphds_edge_callback, for_each_edge): New parameter.
18872 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18874 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
18875 out code checking if runtime alias check is possible to below ...
18876 Call the new function.
18877 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
18878 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
18880 2017-06-07 Marek Polacek <polacek@redhat.com>
18883 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
18884 TYPE_OVERFLOW_WRAPS check.
18886 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18888 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
18889 if versioning is required.
18890 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
18891 peeling with the check for versioning.
18893 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18895 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
18896 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
18897 Set true to new parameter if new ssa variable is defined.
18898 (vect_gen_vector_loop_niters): Refactor. Set range information
18899 for the new vector loop bound variable.
18900 (vect_do_peeling): Ditto.
18902 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18904 * tree-affine.c (ssa.h): Include header file.
18905 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
18906 has wrapping overflow behavior.
18908 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18910 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
18912 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18914 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
18915 (tree_to_aff_combination): ... here.
18917 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18919 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
18920 reg_pressure model function.
18921 (ivopts_global_cost_for_size): Delete.
18922 (determine_set_costs, iv_ca_recount_cost): Call new model function
18923 ivopts_estimate_reg_pressure.
18925 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18927 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
18928 expensive than udiv. Remove floating point cases from mod.
18930 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18932 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
18933 Increase idiv cost.
18935 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18937 * config/aarch64/aarch64.md
18938 (copysignsf3): Fix mask generation.
18940 2017-06-07 Jakub Jelinek <jakub@redhat.com>
18942 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
18944 (class dump_manager): Add register_dumps method.
18945 * dumpfile.c: Include langhooks.h.
18946 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
18947 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
18948 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
18949 (dump_manager::dump_register): Start with 512 entries instead of 32.
18950 (dump_manager::register_dumps): New method.
18951 * toplev.c (general_init): Instead of invoking register_dumps
18952 langhook, invoke register_dumps method on the dump manager.
18953 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
18956 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
18958 * doc/md.texi: Clarify the restrictions on a define_insn condition.
18959 Say that # requires an associated define_split to exist, and that
18960 the define_split must be suitable for use after register allocation.
18962 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18964 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
18965 (compute_outgoing_frequencies): Also initialize zero counts.
18966 (find_many_sub_basic_blocks): Do not produce uninitialized profile
18967 around loops; preserve more of profile when nothing changes.
18969 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
18971 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
18973 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
18974 * config/arm/arm-cpu-cdata.h: Regenerate.
18975 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
18976 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
18977 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
18978 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
18979 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
18981 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
18983 * config/arm/t-rmprofile: Likewise.
18984 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
18986 2017-06-06 David S. Miller <davem@davemloft.net>
18989 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
18990 blockage if function uses alloca.
18992 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18994 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
18995 New "uid" fields to hold pretty-print IDs of group and ref.
18996 Memory references are now identified as <group_id>:<ref_id>
18997 instead of using [random] addresses.
18998 (dump_mem_details): Simplify, no functional change.
18999 (dump_mem_ref): Simplify and make output more concise.
19000 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
19001 (find_or_create_group): Initialize group uid.
19002 (record_ref): Initialize ref uid. Improve debug output.
19003 (prune_group_by_reuse, should_issue_prefetch_p,)
19004 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
19005 (mark_nontemporal_store, determine_loop_nest_reuse):
19006 Improve debug output.
19008 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19010 * dbgcnt.def (prefetch): New debug counter.
19011 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
19012 (schedule_prefetches): Stop issueing prefetches if debug counter
19015 2017-06-06 Tom de Vries <tom@codesourcery.com>
19017 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
19018 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
19020 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19022 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
19023 Use aarch64_reg_or_zero predicate for operand 4.
19024 (aarch64_compare_and_swap<mode> define_insn_and_split):
19025 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
19026 (aarch64_store_exclusive<mode>): Likewise for operand 2.
19028 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19030 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
19031 (arm_compute_save_core_reg_mask): This.
19032 (thumb1_compute_save_reg_mask): Rename into ...
19033 (thumb1_compute_save_core_reg_mask): This.
19034 (arm_compute_save_reg0_reg12_mask): Adapt comment.
19035 (arm_compute_frame_layout): Likewise.
19037 2017-06-06 Richard Biener <rguenther@suse.de>
19039 PR tree-optimization/80974
19040 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
19041 keep or clear leaders SSA info.
19043 2017-06-06 Tom de Vries <tom@codesourcery.com>
19045 * config/nvptx/nvptx.c (split_mode_p): New function.
19046 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
19048 2017-06-06 Tom de Vries <tom@codesourcery.com>
19050 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
19052 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19055 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
19058 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19060 * shrink-wrap.c (handle_simple_exit): Update profile.
19061 (try_shrink_wrapping): Upate profile.
19063 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19065 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
19066 (tree_guess_outgoing_edge_probabilities): New.
19067 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
19068 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
19070 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19072 * ipa-split.c (split_function): Initialize return bb profile.
19074 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19076 * profile.c (compute_branch_probabilities): Also initialize
19077 EXIT_BLOCK profile.
19079 2017-06-06 Richard Biener <rguenther@suse.de>
19081 PR tree-optimization/80928
19082 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
19083 (vect_analyze_loop_operations): Properly guard analysis for
19085 (vect_transform_loop): Likewise.
19086 (vect_analyze_loop_2): Also reset SLP type on PHIs.
19087 (vect_model_induction_cost): Do not cost for pure SLP.
19088 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
19089 of induction in inner loop vectorization.
19090 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
19091 (vect_get_and_check_slp_defs): Handle vect_induction_def.
19092 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
19094 (vect_analyze_slp_cost_1): Cost induction.
19095 (vect_detect_hybrid_slp_stmts): Handle PHIs.
19096 (vect_get_slp_vect_defs): Likewise.
19097 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
19098 (vect_transform_stmt): Handle SLP reductions.
19099 * tree-vectorizer.h (vectorizable_induction): Adjust.
19101 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19103 * config/rs6000/rs6000.c (make_resolver_func): Update
19104 init_lowered_empty_function call.
19106 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
19108 * doc/invoke.texi: Document the -fprofile-abs-path option.
19109 * common.opt (fprofile-abs-path): New option.
19110 * gcov-io.h (gcov_write_filename): Declare.
19111 * gcov-io.c (gcov_write_filename): New function.
19112 * coverage.c (coverage_begin_function): Use gcov_write_filename.
19113 * profile.c (output_location): Likewise.
19115 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19117 * shring-wrap.c: Revert accidental commit.
19119 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
19121 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
19123 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19125 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
19127 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
19128 profile in callgraph edge.
19129 * profile-count.h (apply_probability): If THIS is 0, then result is 0
19130 (apply_scale): Likewise.
19131 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
19132 Also scale profile when inlining function with zero profile.
19133 (initialize_cfun): Update exit block profile even when it is zero.
19134 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
19135 when profile is read.
19137 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19139 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
19140 (CLONE_*): New constants to define the processors we can generate
19141 code for with the target_clone attribute.
19142 (rs6000_clone_map): New array to identify which clone processors
19143 the current program is running on.
19144 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
19145 target_clone attribute.
19146 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
19147 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
19148 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
19149 (cpu_expand_builtin): Add support for target_clone attribute.
19150 (rs6000_valid_attribute_p): Allow "default" attribute.
19151 (get_decl_name): New debug function to simplify printing the
19152 current function name in debugging statements.
19153 (rs6000_clone_priority): New functions to support the target_clone
19154 attribute, and be able to generate code to switch between ISA 2.05
19155 through ISA 3.0 (power6 through power9).
19156 (rs6000_compare_version_priority): Likewise.
19157 (rs6000_get_function_versions_dispatcher): Likewise.
19158 (make_resolver_func): Likewise.
19159 (add_condition_to_bb): Likewise.
19160 (dispatch_function_versions): Likewise.
19161 (rs6000_generate_version_dispatcher_body): Likewise.
19162 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
19163 (fusion_gpr_load_p): Fix a spacing issue.
19164 * doc/extend.texi (Common Function Attributes): Document that the
19165 PowerPC supports the target_clone attribute.
19167 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19169 * config/arm/arm.h: explain F symbol found in description of ARM
19170 register allocation in its legend.
19172 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19174 * config/mips/frame-header-opt.c: Include profile-count.h.
19175 * config/riscv/riscv.c: Include profile-count.h
19177 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19179 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
19181 (sm_set_flag_if_changed): Add bbs field.
19182 (execute_sm_if_changed_flag_set): Pass BBS.
19183 (execute_sm): Update.
19185 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19187 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
19190 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19192 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
19193 (peephole2): New peephole2 to emit the above.
19194 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
19196 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19198 * config/aarch64/aarch64.c (define_peephole2 above
19199 *sub_<shift>_<mode>): New peephole.
19201 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19203 * config/i386/i386.c (make_resolver_func): Update.
19204 * Makefile.in: Add profile-count.h and profile-count.o
19205 * auto-profile.c (afdo_indirect_call): Update to new API.
19206 (afdo_set_bb_count): Update.
19207 (afdo_propagate_edge): Update.
19208 (afdo_propagate_circuit): Update.
19209 (afdo_calculate_branch_prob): Update.
19210 (afdo_annotate_cfg): Update.
19211 * basic-block.h: Include profile-count.h
19212 (struct edge_def): Turn count to profile_count.
19213 (struct basic_block_def): Likewie.
19214 (REG_BR_PROB_BASE): Move to profile-count.h
19215 (RDIV): Move to profile-count.h
19216 * bb-reorder.c (max_entry_count): Turn to profile_count.
19217 (find_traces): Update.
19218 (rotate_loop):Update.
19219 (connect_traces):Update.
19220 (sanitize_hot_paths):Update.
19221 * bt-load.c (migrate_btr_defs): Update.
19222 * cfg.c (RDIV): Remove.
19223 (init_flow): Use alloc_block.
19224 (alloc_block): Uninitialize count.
19225 (unchecked_make_edge): Uninitialize count.
19226 (check_bb_profile): Update.
19227 (dump_edge_info): Update.
19228 (dump_bb_info): Update.
19229 (update_bb_profile_for_threading): Update.
19230 (scale_bbs_frequencies_int): Update.
19231 (scale_bbs_frequencies_gcov_type): Update.
19232 (scale_bbs_frequencies_profile_count): New.
19233 * cfg.h (update_bb_profile_for_threading): Update.
19234 (scale_bbs_frequencies_profile_count): Declare.
19235 * cfgbuild.c (compute_outgoing_frequencies): Update.
19236 (find_many_sub_basic_blocks): Update.
19237 * cfgcleanup.c (try_forward_edges): Update.
19238 (try_crossjump_to_edge): Update.
19239 * cfgexpand.c (expand_gimple_tailcall): Update.
19240 (construct_exit_block): Update.
19241 * cfghooks.c (verify_flow_info): Update.
19242 (dump_bb_for_graph): Update.
19243 (split_edge): Update.
19244 (make_forwarder_block): Update.
19245 (duplicate_block): Update.
19246 (account_profile_record): Update.
19247 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
19248 (get_estimated_loop_iterations): Update.
19249 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
19250 (single_likely_exit): Update.
19251 * cfgloopmanip.c (scale_loop_profile): Update.
19253 (set_zero_probability): Update.
19254 (lv_adjust_loop_entry_edge): Update.
19255 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
19256 (purge_dead_edges): Update.
19257 (rtl_account_profile_record): Update.
19258 * cgraph.c (cgraph_node::create): Uninitialize count.
19259 (symbol_table::create_edge): Uninitialize count.
19260 (cgraph_update_edges_for_call_stmt_node): Update.
19261 (cgraph_edge::dump_edge_flags): Update.
19262 (cgraph_node::dump): Update.
19263 (cgraph_edge::maybe_hot_p): Update.
19264 * cgraph.h: Include profile-count.h
19265 (create_clone), create_edge, create_indirect_edge): Update.
19266 (cgraph_node): Turn count to profile_count.
19267 (cgraph_edge0: Likewise.
19268 (make_speculative, clone): Update.
19269 (create_edge): Update.
19270 (init_lowered_empty_function): Update.
19271 * cgraphclones.c (cgraph_edge::clone): Update.
19272 (duplicate_thunk_for_node): Update.
19273 (cgraph_node::create_clone): Update.
19274 * cgraphunit.c (cgraph_node::analyze): Update.
19275 (cgraph_node::expand_thunk): Update.
19276 * final.c (dump_basic_block_info): Update.
19277 * gimple-streamer-in.c (input_bb): Update.
19278 * gimple-streamer-out.c (output_bb): Update.
19279 * graphite.c (print_global_statistics): Update.
19280 (print_graphite_scop_statistics): Update.
19281 * hsa-brig.c: Include basic-block.h.
19282 * hsa-dump.c: Include basic-block.h.
19283 * hsa-gen.c (T sum_slice): Update.
19284 (convert_switch_statements):Update.
19285 * hsa-regalloc.c: Include basic-block.h.
19286 * ipa-chkp.c (chkp_produce_thunks): Update.
19287 * ipa-cp.c (struct caller_statistics): Update.
19288 (init_caller_stats): Update.
19289 (gather_caller_stats): Update.
19290 (ipcp_cloning_candidate_p): Update.
19291 (good_cloning_opportunity_p): Update.
19292 (get_info_about_necessary_edges): Update.
19293 (dump_profile_updates): Update.
19294 (update_profiling_info): Update.
19295 (update_specialized_profile): Update.
19296 (perhaps_add_new_callers): Update.
19297 (decide_about_value): Update.
19298 (ipa_cp_c_finalize): Update.
19299 * ipa-devirt.c (struct odr_type_warn_count): Update.
19300 (struct decl_warn_count): Update.
19301 (struct final_warning_record): Update.
19302 (possible_polymorphic_call_targets): Update.
19303 (ipa_devirt): Update.
19304 * ipa-fnsummary.c (redirect_to_unreachable): Update.
19305 * ipa-icf.c (sem_function::merge): Update.
19306 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
19307 * ipa-inline.c (compute_uninlined_call_time): Update.
19308 (compute_inlined_call_time): Update.
19309 (want_inline_small_function_p): Update.
19310 (want_inline_self_recursive_call_p): Update.
19311 (edge_badness): Update.
19312 (lookup_recursive_calls): Update.
19313 (recursive_inlining): Update.
19314 (inline_small_functions): Update.
19315 (dump_overall_stats): Update.
19316 (dump_inline_stats): Update.
19317 * ipa-profile.c (ipa_profile_generate_summary): Update.
19318 (ipa_propagate_frequency): Update.
19319 (ipa_profile): Update.
19320 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19321 * ipa-utils.c (ipa_merge_profiles): Update.
19322 * loop-doloop.c (doloop_modify): Update.
19323 * loop-unroll.c (report_unroll): Update.
19324 (unroll_loop_runtime_iterations): Update.
19325 * lto-cgraph.c (lto_output_edge): Update.
19326 (lto_output_node): Update.
19327 (input_node): Update.
19328 (input_edge): Update.
19329 (merge_profile_summaries): Update.
19330 * lto-streamer-in.c (input_cfg): Update.
19331 * lto-streamer-out.c (output_cfg): Update.
19332 * mcf.c (create_fixup_graph): Update.
19333 (adjust_cfg_counts): Update.
19334 (sum_edge_counts): Update.
19335 * modulo-sched.c (sms_schedule): Update.
19336 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
19337 * predict.c (maybe_hot_count_p): Update.
19338 (probably_never_executed): Update.
19339 (dump_prediction): Update.
19340 (combine_predictions_for_bb): Update.
19341 (propagate_freq): Update.
19342 (handle_missing_profiles): Update.
19343 (counts_to_freqs): Update.
19344 (rebuild_frequencies): Update.
19345 (force_edge_cold): Update.
19346 * predict.h: Include profile-count.h
19347 (maybe_hot_count_p, counts_to_freqs): UPdate.
19348 * print-rtl-function.c: Do not include cfg.h
19349 * print-rtl.c: Include basic-block.h
19350 * profile-count.c: New file.
19351 * profile-count.h: New file.
19352 * profile.c (is_edge_inconsistent): Update.
19353 (correct_negative_edge_counts): Update.
19354 (is_inconsistent): Update.
19355 (set_bb_counts): Update.
19356 (read_profile_edge_counts): Update.
19357 (compute_frequency_overlap): Update.
19358 (compute_branch_probabilities): Update; Initialize and deinitialize
19360 (branch_prob): Update.
19361 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
19362 (edge_gcov_count): New.
19363 (bb_gcov_count): New.
19364 * shrink-wrap.c (try_shrink_wrapping): Update.
19365 * tracer.c (better_p): Update.
19366 * trans-mem.c (expand_transaction): Update.
19367 (ipa_tm_insert_irr_call): Update.
19368 (ipa_tm_insert_gettmclone_call): Update.
19369 * tree-call-cdce.c: Update.
19370 * tree-cfg.c (gimple_duplicate_sese_region): Update.
19371 (gimple_duplicate_sese_tail): Update.
19372 (gimple_account_profile_record): Update.
19373 (execute_fixup_cfg): Update.
19374 * tree-inline.c (copy_bb): Update.
19375 (copy_edges_for_bb): Update.
19376 (initialize_cfun): Update.
19377 (freqs_to_counts): Update.
19378 (copy_cfg_body): Update.
19379 (expand_call_inline): Update.
19380 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19381 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
19382 (try_unroll_loop_completely): Update.
19383 (try_peel_loop): Update.
19384 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19385 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
19386 * tree-ssa-loop-split.c (connect_loops): Update.
19387 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
19388 * tree-ssa-reassoc.c (branch_fixup): Update.
19389 * tree-ssa-tail-merge.c (replace_block_by): Update.
19390 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
19391 (compute_path_counts): Update.
19392 (update_profile): Update.
19393 (recompute_probabilities): Update.
19394 (update_joiner_offpath_counts): Update.
19395 (estimated_freqs_path): Update.
19396 (freqs_to_counts_path): Update.
19397 (clear_counts_path): Update.
19398 (ssa_fix_duplicate_block_edges): Update.
19399 (duplicate_thread_path): Update.
19400 * tree-switch-conversion.c (case_bit_test_cmp): Update.
19401 (struct switch_conv_info): Update.
19402 * tree-tailcall.c (decrease_profile): Update.
19403 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19404 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19405 * value-prof.c (check_counter): Update.
19406 (gimple_divmod_fixed_value): Update.
19407 (gimple_mod_pow2): Update.
19408 (gimple_mod_subtract): Update.
19409 (gimple_ic_transform): Update.
19410 (gimple_stringop_fixed_value): Update.
19411 * value-prof.h (gimple_ic): Update.
19413 2017-06-02 Carl Love <cel@us.ibm.com>
19415 * config/rs6000/rs6000-c: Add support for built-in functions
19416 vector double vec_doublee (vector signed int);
19417 vector double vec_doublee (vector unsigned int);
19418 vector double vec_doublee (vector float);
19419 vector double vec_doubleh (vector signed int);
19420 vector double vec_doubleh (vector unsigned int);
19421 vector double vec_doubleh (vector float);
19422 vector double vec_doublel (vector signed int);
19423 vector double vec_doublel (vector unsigned int);
19424 vector double vec_doublel (vector float);
19425 vector double vec_doubleo (vector signed int);
19426 vector double vec_doubleo (vector unsigned int);
19427 vector double vec_doubleo (vector float);.
19428 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
19429 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
19431 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
19432 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
19433 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
19435 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
19436 vec_doublel, vec_doubleh.
19437 * doc/extend.texi: Update the built-in documentation file for the
19438 new built-in functions.
19440 2017-06-02 David Malcolm <dmalcolm@redhat.com>
19443 * ipa-inline-analysis.c (free_growth_caches): Set
19444 edge_removal_hook_holder to NULL after removing it.
19446 2017-06-02 Sudakshina Das <sudi.das@arm.com>
19448 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
19449 comparision with zero.
19451 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19452 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19453 for early expansion of vec_min and vec_max builtins.
19454 (builtin_function_type): Add min/max unsigned variants to those
19455 identified as having unsigned arguments.
19457 2017-06-02 Olivier Hainque <hainque@adacore.com>
19459 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
19461 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19463 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
19464 Use VALL_F16 iterator rather than VALL.
19466 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19468 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
19469 Emit CBNZ inside loop when doing a strong exchange and comparing
19470 against zero. Generate the CC flags after the loop.
19472 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
19474 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
19475 (dl_section_ref): New.
19476 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
19477 On AIX, append an expression to subtract the size of the
19478 section length to dl_section_ref.
19480 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19482 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19483 for early expansion of vector absolute builtins.
19485 2017-06-02 Richard Biener <rguenther@suse.de>
19487 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
19488 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
19490 2017-06-02 Richard Biener <rguenther@suse.de>
19492 PR tree-optimization/80948
19493 * tree-tailcall.c (find_tail_calls): Track stmts to move in
19494 stmt order as well.
19496 2017-06-02 Richard Biener <rguenther@suse.de>
19498 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
19500 * tree-vect-stmts.c (process_use): Do not mark backedge defs
19501 for inductions as relevant.
19503 2017-06-02 Richard Biener <rguenther@suse.de>
19505 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
19506 (vectorizable_induction): ... this. Remove dead code.
19508 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
19510 * builtins. (expand_builtin_alloca): Remove second parameter and
19511 infer its value from the first parameter instead.
19512 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
19514 2017-06-02 Jakub Jelinek <jakub@redhat.com>
19516 PR rtl-optimization/80903
19517 * loop-doloop.c (add_test): Unshare sequence.
19519 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
19521 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
19523 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
19525 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
19527 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
19528 xlogue_layout::get_instance, logue_layout::xlogue_layout,
19529 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
19530 (xlogue_layout::get_stub_rtx): Make static.
19531 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
19532 (xlogue_layout::compute_stub_managed_regs): Rename to...
19533 (xlogue_layout::count_stub_managed_regs): ...this.
19534 (xlogue_layout::is_stub_managed_reg): New function.
19535 (xlogue_layout::m_stub_names): Rename to...
19536 (xlogue_layout::s_stub_names): ...this, make static.
19537 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
19538 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
19539 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
19540 xlogue_layout::s_stub_names): Instantiate statics.
19541 (stub_managed_regs): Remove.
19542 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
19543 (disable_call_ms2sysv_xlogues): Rename to...
19544 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
19545 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
19547 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
19548 change after reload_completed.
19549 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
19551 (ix86_expand_prologue): Likewise.
19552 (ix86_expand_epilogue): Likewise.
19553 (ix86_expand_split_stack_prologue): Likewise.
19554 (ix86_compute_frame_layout): Remove frame parameter ...
19555 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
19556 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
19558 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
19559 (ix86_frame): Move from here ...
19560 * config/i386/i386.h (ix86_frame): ... to here.
19561 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
19562 complete ix86_frame data structure instead. Remove some_ld_name.
19564 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19566 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
19567 symbols that hold a DECL_VALUE_EXPR.
19569 2017-06-01 Martin Jambor <mjambor@suse.cz>
19571 PR tree-optimization/80898
19572 * tree-sra.c (process_subtree_disqualification): Removed.
19573 (disqualify_candidate): Do not acll
19574 process_subtree_disqualification.
19575 (subtree_mark_written_and_enqueue): New function.
19576 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
19577 RHS has been disqualified and re-queue LHS if necessary. Apart
19578 from that, ignore disqualified RHS.
19580 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19582 * config/s390/s390.c (s390_emit_epilogue): Disable early return
19583 address fetch for z10 or later.
19585 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19587 * config/arc/arc.md (tst_movb): Add guard when splitting.
19589 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19591 * config/arc/arc.c (arc_can_eliminate): Test against
19592 arc_frame_pointer_needed.
19594 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19596 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
19597 to prevent store reordering.
19598 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
19599 (type): Add block type.
19600 (stack_tie): Define special instruction to be used in
19603 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19605 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
19606 constraint. It is not valid for the pattern.
19607 (noncommutative_binary_comparison): Likewise.
19609 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19611 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
19614 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19616 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
19617 be used by the reg-alloc.
19619 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19621 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
19622 reg-alloc when having mul64 or mul32x16 instructions.
19623 (mulsidi3): Likewise.
19624 (umulsidi3): Likewise.
19625 (mulsi32x16): New pattern.
19626 (mulsi64): Likewise.
19627 (mulsidi64): Likewise.
19628 (umulsidi64): Likewise.
19629 (MUL32x16_REG): Define.
19630 (mul64_600): Use MUL32x16_REG.
19631 (mac64_600): Likewise.
19632 (umul64_600): Likewise.
19633 (umac64_600): Likewise.
19635 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19637 * config/arc/arc.md (mulsi3_700): Make it commutative.
19639 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
19641 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
19643 (*sign_extendsidi2_insn): Likewise for movstosw.
19645 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19647 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
19648 the type of the input discriminant value. Convert the
19649 discriminant value of signedness vary.
19651 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
19653 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
19654 Add to -Wall section.
19656 2017-06-01 Richard Biener <rguenther@suse.de>
19658 PR middle-end/66313
19659 * fold-const.c (fold_plusminus_mult_expr): If the factored
19660 factor may be zero use a wrapping type for the inner operation.
19661 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
19662 and handle moved defs.
19663 (process_assignment): Properly guard the unary op case. Return a
19664 tri-state indicating that moving the stmt before the call may allow
19665 to continue. Pass through to_move.
19666 (find_tail_calls): Handle moving unrelated defs before
19669 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
19672 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
19673 splitter result in the canonical way.
19675 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
19677 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
19678 also for 32bit target. Update insn attributes.
19679 (zero-extendsidi2 splitter): Allow all registers for operand 1.
19681 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
19683 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
19684 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
19685 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
19686 (_mm_maskz_min_ss): New intrinsics.
19688 2017-05-31 Martin Liska <mliska@suse.cz>
19690 * tree-vect-loop.c (vect_create_epilog_for_reduction):
19691 Change comment style to one we normally use.
19692 (vectorizable_reduction): Likewise.
19693 (vectorizable_induction): Likewise.
19694 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19695 (vectorizable_call): Likewise.
19696 (vectorizable_simd_clone_call): Likewise.
19697 (vectorizable_conversion): Likewise.
19698 (vectorizable_assignment): Likewise.
19699 (vectorizable_shift): Likewise.
19700 (vectorizable_operation): Likewise.
19701 (vectorizable_store): Likewise.
19702 (vectorizable_load): Likewise.
19703 * tree-vectorizer.h: Likewise.
19705 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
19707 * passes.c (emergency_dump_function): New.
19708 * tree-pass.h (emergency_dump_function): Declare.
19709 * plugin.c (plugins_internal_error_function): Remove.
19710 * plugin.h (plugins_internal_error_function): Remove declaration.
19711 * toplev.c (internal_error_function): New static function. Use it...
19712 (general_init): ...here.
19714 2017-05-31 Graham Markall <graham.markall@embecosm.com>
19716 * config/arc/arc.c (arc_print_operand): Handle constant operands.
19717 (arc_rtx_costs): Add costs for new patterns.
19718 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
19719 * config/arc/predicates.md: Add _1_2_3_operand predicate.
19721 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19723 * tree-ssa-strlen.c (get_next_strinfo): New function.
19724 (get_stridx_plus_constant): Use it.
19725 (zero_length_string): Likewise.
19726 (adjust_related_strinfos): Likewise.
19727 (adjust_last_stmt): Likewise.
19729 2017-05-31 Richard Biener <rguenther@suse.de>
19732 * config/i386/i386.c (ix86_expand_builtin): Remove assert
19733 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
19735 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19737 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
19738 loop_vinfo argument and use of dependence distance vectors.
19739 Check instead whether the two references differ only in their
19740 initial value and assume that they have the same alignment if the
19741 difference is a multiple of the vector alignment.
19742 (vect_analyze_data_refs_alignment): Update call accordingly.
19744 2017-05-31 Martin Liska <mliska@suse.cz>
19747 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
19749 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19751 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
19752 (create_intersect_range_checks): Move from ...
19753 * tree-data-ref.c (create_intersect_range_checks_index)
19754 (create_intersect_range_checks): ... to here.
19755 (create_runtime_alias_checks): New function factored from ...
19756 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
19757 here. Call above function.
19758 * tree-data-ref.h (create_runtime_alias_checks): New function.
19760 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19762 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
19763 segment length for dr_b and compute it in wide_int.
19765 2017-05-31 Richard Biener <rguenther@suse.de>
19767 PR tree-optimization/80906
19768 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
19769 and pass through iv_map.
19770 (copy_bb_and_scalar_dependences): Adjust.
19771 (translate_pending_phi_nodes): Likewise.
19772 (copy_loop_close_phi_args): Handle code-generating IVs instead
19775 2017-05-30 David Malcolm <dmalcolm@redhat.com>
19777 * diagnostic-color.c (color_dict): Add "type-diff".
19778 (parse_gcc_colors): Update comment.
19779 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
19780 -fdiagnostics-show-template-tree and -fno-elide-type.
19781 (GCC_COLORS): Add type-diff to example.
19783 (-fdiagnostics-show-template-tree): New.
19784 (-fno-elide-type): New.
19785 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
19786 the pp_format_decoder callback. Call any m_format_postprocessor's
19788 (pretty_printer::pretty_printer): Initialize
19789 m_format_postprocessor.
19790 (pretty_printer::~pretty_printer): Delete any
19791 m_format_postprocessor.
19792 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
19793 (class format_postprocessor): New class.
19794 (struct pretty_printer::format_decoder): Document the new parameters.
19795 (struct pretty_printer::m_format_postprocessor): New field.
19796 * tree-diagnostic.c (default_tree_printer): Update for new
19797 bool and const char ** params.
19798 * tree-diagnostic.h (default_tree_printer): Likewise.
19800 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
19802 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
19803 (lwa_operand): Delete rs6000_gen_cell_microcode test.
19804 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19805 rs6000_gen_cell_microcode code.
19806 (rs6000_final_prescan_insn): Delete.
19807 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
19808 "warn-cell-microcode" entries.
19809 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
19810 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
19811 throughout. Change cc_reg_not_micro_cr0_operand to
19812 cc_reg_not_cr0_operand throughout.
19813 (*extendhi<mode>2_noload): Delete.
19814 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
19815 (mwarn-cell-microcode): Delete.
19816 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
19817 -mgen-cell-microcode and -mwarn-cell-microcode.
19819 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
19822 * config/i386/constraints.md (Yd): New constraint.
19824 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
19825 and (?Yd, r) alternatives. Update insn attributes.
19826 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
19827 and (?*Yd, r) alternatives. Update insn attributes.
19828 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
19830 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
19832 * gimplify.c (gimplify_modify_expr): Don't create a
19833 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
19836 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
19838 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
19840 2017-05-30 Richard Biener <rguenther@suse.de>
19842 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
19843 and reduc_def fields.
19844 (STMT_VINFO_REDUC_TYPE): New define.
19845 (STMT_VINFO_REDUC_DEF): Likewise.
19846 (vect_force_simple_reduction): Adjust prototype.
19847 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
19848 (vect_is_simple_reduction): Remove check_reduction argument.
19849 (vect_force_simple_reduction): Adjust and set
19850 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19851 (vectorizable_reduction): Do not re-do reduction analysis
19852 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19853 * tree-parloops.c (gather_scalar_reductions): Adjust.
19855 2017-05-30 Richard Biener <rguenther@suse.de>
19857 PR middle-end/80901
19858 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
19861 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
19863 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
19864 Introduce unknown_misalignment parameter and remove vf.
19865 (vect_peeling_hash_get_lowest_cost):
19866 Pass unknown_misalignment parameter.
19867 (vect_enhance_data_refs_alignment):
19868 Fix unsupportable data ref treatment.
19870 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19872 * tree-vect-data-refs.c (vect_get_data_access_cost):
19873 Workaround for SLP handling.
19874 (vect_enhance_data_refs_alignment):
19875 Compute costs for doing no peeling at all, compare to the best
19876 peeling costs so far and avoid peeling if cheaper.
19878 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19880 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
19881 Return peeling info and set costs to zero for unlimited cost
19883 (vect_enhance_data_refs_alignment): Also inspect all datarefs
19884 with unknown misalignment. Compute and costs for unknown
19885 misalignment, compare them to the costs for known misalignment
19886 and choose the cheapest for peeling.
19888 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19890 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
19891 (vect_get_peeling_costs_all_drs): Create function.
19892 (vect_peeling_hash_get_lowest_cost):
19893 Use vect_get_peeling_costs_all_drs.
19894 (vect_peeling_supportable): Create function.
19895 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
19897 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19899 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
19900 DR_HAS_NEGATIVE_STEP.
19901 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
19902 (vect_enhance_data_refs_alignment): Use.
19903 (vect_duplicate_ssa_name_ptr_info): Use.
19904 * tree-vectorizer.h (dr_misalignment): Use.
19905 (known_alignment_for_access_p): Use.
19907 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19910 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
19912 (has_section_name): New function.
19914 2017-05-30 Martin Liska <mliska@suse.cz>
19917 * auto-profile.c (get_function_decl_from_block): Fix
19920 2017-05-30 Richard Biener <rguenther@suse.de>
19922 PR middle-end/80876
19923 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
19925 2017-05-30 Martin Liska <mliska@suse.cz>
19927 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
19928 * dumpfile.h (struct dump_file_info): Remove ctors.
19930 2017-05-30 Martin Liska <mliska@suse.cz>
19932 * predict.def: Fix GNU coding style.
19934 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
19936 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19937 Mark 'to' argument with ATTRIBUTE_UNUSED.
19939 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
19941 * config/xtensa/xtensa.c (xtensa_emit_call): Use
19942 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
19943 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
19946 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
19948 * doc/install.texi (Options specification): Restore entry of
19949 --enable-sjlj-exceptions.
19951 2017-05-27 Michael Eager <eager@eagercon.com>
19954 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
19956 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
19958 * config/microblaze/microblaze.h
19959 (FIXED_REGISTERS): Update in macro.
19960 (CALL_USED_REGISTERS): Update in macro.
19962 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
19964 * doc/install.texi: Add links to macOS binary distributions.
19966 2017-05-27 Jakub Jelinek <jakub@redhat.com>
19970 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19972 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
19974 2017-05-26 Martin Liska <mliska@suse.cz>
19976 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
19978 2017-05-26 Martin Liska <mliska@suse.cz>
19980 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
19981 always leading ';; '.
19982 (dump_bb_info): Likewise.
19983 (brief_dump_cfg): Likewise.
19984 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
19985 * dumpfile.c: Remove usage of TDF_VERBOSE.
19986 * dumpfile.h (enum dump_kind): Likewise.
19987 (dump_gimple_bb_header): Do not use TDF_COMMENT.
19988 * print-tree.c (debug_verbose): Remove.
19989 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
19990 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
19991 * tree-diagnostic.c (default_tree_printer): Replace
19992 TDF_DIAGNOSTIC with TDF_SLIM.
19994 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19996 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
19997 in parameter loop, rather than loop_vinfo.
19998 (create_intersect_range_checks): Ditto.
19999 (vect_create_cond_for_alias_checks): Update call to above functions.
20001 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20003 PR tree-optimization/80815
20004 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
20005 for merging runtime alias checks. Handle negative DR_STEPs.
20007 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20009 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
20011 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
20012 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
20013 out code pruning runtime alias checks.
20014 * tree-data-ref.c (prune_runtime_alias_test_list): New function
20015 factored out from above.
20016 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20018 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20020 (prune_runtime_alias_test_list): New decalaration.
20022 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20024 * tree-vect-data-refs.c (compare_tree): Rename and move ...
20025 * tree-data-ref.c (data_ref_compare_tree): ... to here.
20026 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
20027 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
20028 (operator==, comp_dr_with_seg_len_pair): Ditto.
20029 (vect_prune_runtime_alias_test_list): Ditto.
20031 2017-05-26 Martin Liska <mliska@suse.cz>
20034 * params.def: Bound partial-inlining-entry-probability param.
20036 2017-05-26 Marek Polacek <polacek@redhat.com>
20039 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
20042 2017-05-26 Richard Biener <rguenther@suse.de>
20044 PR tree-optimization/80842
20045 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
20048 2017-05-26 Richard Biener <rguenther@suse.de>
20050 PR tree-optimization/80844
20051 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
20053 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
20055 * doc/md.texi (Machine Constraints): Update x86 family
20056 machine constraints section to match 'config/i386/constraints.md'.
20058 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
20060 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
20062 2017-05-25 Nathan Sidwell <nathan@acm.org>
20064 * doc/invoke.texi (--enable-languages): Update documentation.
20066 2017-05-25 Martin Liska <mliska@suse.cz>
20068 * dumpfile.c: Add TDF_FOLDING.
20069 * dumpfile.h (enum dump_kind): Likewise.
20070 * genmatch.c (dt_simplify::gen_1): Use it.
20072 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20074 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
20076 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20078 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20079 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
20081 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20083 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
20084 * match.pd (X == C): Rewrite it here.
20085 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
20086 with_certain_nonzero_bits2): New predicates.
20087 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
20089 2017-05-24 Nathan Sidwell <nathan@acm.org>
20091 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
20094 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
20097 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
20099 * config/powerpcspe: New port. Files are copied from the rs6000
20100 port, with "rs6000" in filenames replaced by "powerpcspe".
20102 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
20104 PR rtl-optimization/80754
20105 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
20107 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
20109 * config/sparc/sparc.md (length): Return the correct value for -mflat
20110 sibcalls to match output_sibcall.
20112 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
20116 * config/rs6000/rs6000.c (struct machine_function): Add new field
20118 (rs6000_get_separate_components): Init that field, use it.
20119 (rs6000_components_for_bb): Use the field.
20121 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
20123 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
20125 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
20127 PR middle-end/80823
20128 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
20130 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20133 * config/s390/s390.c (s390_check_qrst_address): Check incoming
20134 address against address_operand predicate.
20135 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
20137 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
20139 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
20140 parameters passed indirectly.
20142 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
20144 * config/i386/i386.md (*movdi_internal): Remove SSE4
20145 alternative 18 (?r, *v). Update insn attributes.
20146 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
20147 Update insn attributes.
20148 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
20149 Update insn attributes.
20150 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
20151 alternative 1 (r, v). Remove isa attribute.
20152 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
20153 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
20154 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
20156 2017-05-23 Tom de Vries <tom@codesourcery.com>
20158 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
20161 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20163 * cgraphunit.c (symbol_table::process_new_functions): Update.
20164 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
20165 (inline_generate_summary): Rename to ...
20166 (ipa_fn_summary_generate): ... this one.
20167 (inline_read_summary): Rename to ...
20168 (ipa_fn_summary_read): ... this one.
20169 (inline_write_summary): Rename to ...
20170 (ipa_fn_summary_write): ... this one.
20171 (inline_free_summary): Rename to ...
20172 (ipa_free_fn_summary): ... this one.
20173 (pass_data_local_fn_summary, pass_local_fn_summary,
20174 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
20175 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
20176 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
20177 make_pass_ipa_fn_summary): New.
20178 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
20179 inline_write_summary, inline_free_summary): Remove.
20180 (ipa_free_fn_summary) : New.
20181 * ipa-inline.c (ipa_inline): Update.
20182 (pass_ipa_inline): Do not generate summaries.
20183 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
20185 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
20186 and add pass_ipa_fn_summary.
20187 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
20189 (make_pass_inline_parameters): Remove.
20191 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
20193 * omp-low.c (struct omp_context): Remove "default_kind" member.
20196 * omp-offload.c (execute_oacc_device_lower): Remove the
20197 parallelism dimensions function attributes for unparallelized
20198 OpenACC kernels constructs.
20200 2017-05-23 Martin Liska <mliska@suse.cz>
20202 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
20204 (cgraph_edge::make_speculative): Likewise.
20205 (cgraph_edge::resolve_speculation): Likewise.
20206 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
20207 (cgraph_node::dump): Likewise.
20208 * cgraph.h: Likewise.
20209 * cgraphunit.c (analyze_functions): Likewise.
20210 (symbol_table::compile): Likewise.
20211 * ipa-cp.c (print_all_lattices): Likewise.
20212 (determine_versionability): Likewise.
20213 (initialize_node_lattices): Likewise.
20214 (ipcp_verify_propagated_values): Likewise.
20215 (estimate_local_effects): Likewise.
20216 (update_profiling_info): Likewise.
20217 (create_specialized_node): Likewise.
20218 (perhaps_add_new_callers): Likewise.
20219 (decide_about_value): Likewise.
20220 (decide_whether_version_node): Likewise.
20221 (identify_dead_nodes): Likewise.
20222 (ipcp_store_bits_results): Likewise.
20223 * ipa-devirt.c (dump_targets): Likewise.
20224 (ipa_devirt): Likewise.
20225 * ipa-icf.c (sem_item::dump): Likewise.
20226 (sem_function::equals): Likewise.
20227 (sem_variable::equals): Likewise.
20228 (sem_item_optimizer::read_section): Likewise.
20229 (sem_item_optimizer::execute): Likewise.
20230 (congruence_class::dump): Likewise.
20231 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
20232 (dump_inline_summary): Likewise.
20233 (estimate_node_size_and_time): Likewise.
20234 (inline_analyze_function): Likewise.
20235 * ipa-inline-transform.c (inline_call): Likewise.
20236 * ipa-inline.c (report_inline_failed_reason): Likewise.
20237 (want_early_inline_function_p): Likewise.
20238 (edge_badness): Likewise.
20239 (update_edge_key): Likewise.
20240 (inline_small_functions): Likewise.
20241 * ipa-profile.c (ipa_profile): Likewise.
20242 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
20243 (ipa_make_edge_direct_to_target): Likewise.
20244 (remove_described_reference): Likewise.
20245 (ipa_impossible_devirt_target): Likewise.
20246 (propagate_controlled_uses): Likewise.
20247 (ipa_print_node_params): Likewise.
20248 (ipcp_transform_function): Likewise.
20249 * ipa-pure-const.c (pure_const_read_summary): Likewise.
20250 (propagate_pure_const): Likewise.
20251 * ipa-reference.c (generate_summary): Likewise.
20252 (read_write_all_from_decl): Likewise.
20253 (propagate): Likewise.
20254 (ipa_reference_read_optimization_summary): Likewise.
20255 * ipa-utils.c (ipa_merge_profiles): Likewise.
20256 * ipa.c (walk_polymorphic_call_targets): Likewise.
20257 (symbol_table::remove_unreachable_nodes): Likewise.
20258 (ipa_single_use): Likewise.
20259 * passes.c (execute_todo): Likewise.
20260 * predict.c (drop_profile): Likewise.
20261 * symtab.c (symtab_node::get_dump_name): New function.
20262 (symtab_node::dump_name): Likewise.
20263 (symtab_node::dump_asm_name): Likewise.
20264 (symtab_node::dump_references): Likewise.
20265 (symtab_node::dump_referring): Likewise.
20266 (symtab_node::dump_base): Likewise.
20267 (symtab_node::debug_symtab): Likewise.
20268 * tree-sra.c (convert_callers_for_node): Likewise.
20269 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
20270 * value-prof.c (init_node_map): Likewise.
20272 2017-05-23 Martin Liska <mliska@suse.cz>
20274 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
20275 and symtab_node::debug_symtab to symbol_table::debug.
20276 * cgraphunit.c (analyze_functions): Use the renamed function.
20277 (symbol_table::compile): Likewise.
20278 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
20279 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
20280 * passes.c (execute_todo): Likewise.
20281 * symtab.c (symbol_table::dump): New function.
20282 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
20284 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20286 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
20287 that nonconst implies exec.
20289 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20291 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
20292 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20293 (inline_edge_summary_vec): Turn into ...
20294 (ipa_call_summaries): ... this one.
20295 (redirect_to_unreachable, edge_set_predicate,
20296 evaluate_properties_for_edge, inline_summary_alloc,
20297 reset_ipa_call_summary, reset_inline_summary,
20298 inline_summary_t::duplicate): Update.
20299 (inline_edge_duplication_hook): Turn to ...
20300 (ipa_call_summary_t::duplicate): ... this one.
20301 (inline_edge_removal_hook): Turn to ...
20302 (ipa_call_summary_t::remove): ... this one.
20303 (dump_inline_edge_summary): Turn to ...
20304 (dump_ipa_call_summary): ... this one.
20305 (estimate_function_body_sizes): Update.
20306 (inline_update_callee_summaries): Update.
20307 (remap_edge_change_prob): Update.
20308 (remap_edge_summaries): Update.
20309 (inline_merge_summary): Update.
20310 (do_estimate_edge_time): Update.
20311 (inline_generate_summary): Update.
20312 (inline_read_section): Update.
20313 (inline_read_summary): Update.
20314 (inline_free_summary): Update.
20315 * ipa-inline.c (can_inline_edge_p): Update.
20316 (compute_inlined_call_time): Update.
20317 (want_inline_small_function_p): Update.
20318 (edge_badness): Update.
20319 (early_inliner): Update.
20320 * ipa-inline.h (inline_edge_summary): Turn to ...
20321 (ipa_call_summary): ... this one.
20322 (ipa_call_summary_t): New class.
20323 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20324 (ipa_call_summaries): New.
20325 (inline_edge_summary): Remove.
20326 (estimate_edge_growth): Update.
20327 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20328 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20329 * ipa-split.c (execute_split_functions): Update.
20330 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20332 2017-05-23 Tom de Vries <tom@codesourcery.com>
20334 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
20335 attributes): Document rdrand effective target.
20337 2017-05-23 Tom de Vries <tom@codesourcery.com>
20339 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
20340 attributes): Sort alphabetically.
20342 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
20344 * config/avr/genmultilib.awk: Use gsub instead of gensub.
20346 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
20349 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
20350 V2DF/V2DI splat into two separate patterns, one that handles
20351 registers, and the other that only handles memory. Drop support
20352 for splatting from a GPR on ISA 2.07 and then splitting the
20353 splat into direct move and splat.
20354 (vsx_splat_<mode>_reg): Likewise.
20355 (vsx_splat_<mode>_mem): Likewise.
20357 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
20359 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
20361 2017-05-22 Jakub Jelinek <jakub@redhat.com>
20363 PR middle-end/80809
20364 * omp-low.c (finish_taskreg_remap): New function.
20365 (finish_taskreg_scan): If unit size of ctx->record_type
20366 is non-constant, unshare the size expression and replace
20367 decls in it with possible outer var refs.
20369 PR middle-end/80809
20370 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
20371 GOVD_SHARED rather than GOVD_PRIVATE with it.
20372 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
20373 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
20375 PR middle-end/80853
20376 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
20377 as last argument to build_outer_var_ref for pointer bases of array
20378 section reductions.
20380 2017-05-19 Martin Sebor <msebor@redhat.com>
20382 * print-tree.c (print_node): Print DECL_READ_P flag.
20384 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20386 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
20387 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
20388 * cgraph.c: Likewise.
20389 * cgraphunit.c: Likewise.
20390 * gengtype.c: Likewise.
20391 * ipa-cp.c: Likewise.
20392 * ipa-devirt.c: Likewise.
20393 * ipa-icf.c: Likewise.
20394 * ipa-predicate.c: Likewise.
20395 * ipa-profile.c: Likewise.
20396 * ipa-prop.c: Likewise.
20397 * ipa-split.c: Likewise.
20399 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
20400 edge_predicate_pool, dump_inline_hints,
20401 inline_summary::account_size_time, redirect_to_unreachable,
20402 edge_set_predicate, set_hint_predicate,
20403 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
20404 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
20405 inline_summary_t::remove, remap_hint_predicate_after_duplication,
20406 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
20407 ipa_call_summary_t::remove, initialize_growth_caches,
20408 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
20409 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
20410 mark_modified, unmodified_parm_1, unmodified_parm,
20411 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
20412 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
20413 compute_bb_predicates, will_be_nonconstant_expr_predicate,
20414 will_be_nonconstant_predicate, record_modified_bb_info,
20415 get_minimal_bb, record_modified, param_change_prob,
20416 phi_result_unknown_predicate, predicate_for_phi_result,
20417 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
20418 estimate_function_body_sizes, compute_inline_parameters,
20419 compute_inline_parameters_for_curren, pass_data_inline_parameters,
20420 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
20421 inline_update_callee_summaries, remap_edge_change_prob,
20422 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
20423 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
20424 inline_analyze_function, inline_summary_t::insert,
20425 inline_generate_summary, read_ipa_call_summary, inline_read_section,
20426 inline_read_summary, write_ipa_call_summary, inline_write_summary,
20427 inline_free_summary): Move to ipa-fnsummary.h
20428 (predicate_t): Remove.
20429 * ipa-fnsummary.c: New file.
20430 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
20431 (enum inline_hints_vals, inline_hints, agg_position_info,
20432 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
20433 inline_summaries, ipa_call_summary, ipa_call_summary_t,
20434 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
20435 dump_inline_summary, dump_inline_hints, inline_generate_summary,
20436 inline_read_summary, inline_write_summary, inline_free_summary,
20437 inline_analyze_function, initialize_inline_failed,
20438 inline_merge_summary, inline_update_overall_summary,
20439 compute_inline_parameters): Move to ipa-fnsummary.h
20440 * ipa-fnsummary.h: New file.
20441 * ipa-inline-transform.h: Include ipa-inline.h.
20442 * ipa-inline.c: LIkewise.
20444 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20446 * ipa-inline.c (edge_badness): Use inlined_time instead of
20447 inline_summaries->get.
20449 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20451 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
20453 2017-05-22 Nathan Sidwell <nathan@acm.org>
20455 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
20456 (fdump-lang): Document 'raw' option.
20457 * dumpfile.h (TDI_tu): Delete.
20458 * dumpfile.c (dump_files): Remove translation-unit.
20459 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
20461 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
20463 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
20464 command option from $(AWK) call.
20465 * config/avr/genmultilib.awk: Simplify and rewrite so that it
20466 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
20467 [FORMAT]: Remove handling of variable.
20468 * config/avr/t-multilib: Regenerate.
20470 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20472 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
20474 (dump_inline_summary): Do not print self_time.
20475 (estimate_function_body_sizes): Do not set self_time.
20476 (compute_inline_parameters): Likewise.
20477 (inline_read_section, inline_write_summary): Do not stream self_time.
20478 * ipa-inline.h (inline_summary): Drop self_time.
20480 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20482 * ipa-inline-analysis.c (account_size_time): Rename to ...
20483 (inline_summary::account_size_time): ... this one.
20484 (reset_ipa_call_summary): Turn to ...
20485 (ipa_call_summary::reset): ... this one.
20486 (reset_inline_summary): Turn to ...
20487 (inline_summary::reset): ... this one.
20488 (inline_summary_t::remove): Update.
20489 (inline_summary_t::duplicate): Update.
20490 (ipa_call_summary_t::remove): Update.
20491 (dump_inline_summary): Update.
20492 (estimate_function_body_sizes): Update.
20493 (compute_inline_parameters): Update.
20494 (estimate_node_size_and_time): Update.
20495 (inline_merge_summary): Update.
20496 (inline_update_overall_summary): Update.
20497 (inline_read_section): Update.
20498 (inline_write_summary): Update.
20499 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
20500 add account_size_time and reset member functions.
20501 (ipa_call_summary): Add reset function.
20502 * ipa-predicate.h (predicate::operator &): Constify.
20504 2017-05-22 Richard Biener <rguenther@suse.de>
20506 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
20508 2017-05-19 Jason Merrill <jason@redhat.com>
20510 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
20512 2017-05-19 Marek Polacek <polacek@redhat.com>
20515 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
20516 TYPE_OVERFLOW_WRAPS checks.
20518 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
20520 * tree-core.h (enum omp_clause_default_kind): Add
20521 "OMP_CLAUSE_DEFAULT_PRESENT".
20522 * tree-pretty-print.c (dump_omp_clause): Handle it.
20523 * gimplify.c (enum gimplify_omp_var_data): Add
20524 "GOVD_MAP_FORCE_PRESENT".
20525 (gimplify_adjust_omp_clauses_1): Map it to
20526 "GOMP_MAP_FORCE_PRESENT".
20527 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
20529 * gimplify.c (oacc_default_clause): Clarify.
20531 2017-05-19 Nathan Sidwell <nathan@acm.org>
20533 LANG_HOOK_REGISTER_DUMPS
20534 * toplev.c (general_init): Call register dump lang hook.
20535 * doc/invoke.texi: Document -fdump-lang option family.
20536 * dumpfile.c (dump_files): Remove class dump here.
20537 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
20538 * dumpfile.h (tree_dump_index): Remove TDI_class.
20539 * langhooks-def.h (lhd_register_dumps): Declare.
20540 (LANG_HOOKS_REGISTER_DUMPS): Define.
20541 (LANG_HOOKS_INITIALIZER): Add it.
20542 * langhooks.c (lhd_register_dumps): Define.
20543 * langhooks.h (struct lang_hooks): Add register_dumps.
20545 2017-05-19 Nathan Sidwell <nathan@acm.org>
20547 * context.h (context::set_passes): New.
20548 * context.c (context::context): Do not create pass manager.
20549 * toplev.c (general_init): Create pass manager here.
20551 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
20553 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
20554 use this splitter if two add or or instructions would also work for
20555 the constant we want to generate.
20557 2017-05-19 Richard Biener <rguenther@suse.de>
20560 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
20561 predicate evaluation.
20563 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20565 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
20567 * ipa-inline.c (want_inline_small_function_p): Do not cast to
20570 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20572 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
20573 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20574 (inline_edge_summary_vec): Turn into ...
20575 (ipa_call_summaries): ... this one.
20576 (redirect_to_unreachable, edge_set_predicate,
20577 evaluate_properties_for_edge, inline_summary_alloc,
20578 reset_ipa_call_summary, reset_inline_summary,
20579 inline_summary_t::duplicate): Update.
20580 (inline_edge_duplication_hook): Turn to ...
20581 (ipa_call_summary_t::duplicate): ... this one.
20582 (inline_edge_removal_hook): Turn to ...
20583 (ipa_call_summary_t::remove): ... this one.
20584 (dump_inline_edge_summary): Turn to ...
20585 (dump_ipa_call_summary): ... this one.
20586 (estimate_function_body_sizes): Update.
20587 (inline_update_callee_summaries): Update.
20588 (remap_edge_change_prob): Update.
20589 (remap_edge_summaries): Update.
20590 (inline_merge_summary): Update.
20591 (do_estimate_edge_time): Update.
20592 (inline_generate_summary): Update.
20593 (inline_read_section): Update.
20594 (inline_read_summary): Update.
20595 (inline_free_summary): Update.
20596 * ipa-inline.c (can_inline_edge_p): Update.
20597 (compute_inlined_call_time): Update.
20598 (want_inline_small_function_p): Update.
20599 (edge_badness): Update.
20600 (early_inliner): Update.
20601 * ipa-inline.h (inline_edge_summary): Turn to ...
20602 (ipa_call_summary): ... this one.
20603 (ipa_call_summary_t): New class.
20604 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20605 (ipa_call_summaries): New.
20606 (inline_edge_summary): Remove.
20607 (estimate_edge_growth): Update.
20608 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20609 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20610 * ipa-split.c (execute_split_functions): Update.
20611 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20613 2017-05-19 Richard Biener <rguenther@suse.de>
20615 PR middle-end/80764
20616 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
20618 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
20620 * config/rs6000/rs6000.c (struct machine_function): Add field
20621 fpr_is_wrapped_separately.
20622 (rs6000_get_separate_components): Use 64 components. Handle the
20623 new FPR components.
20624 (rs6000_components_for_bb): Handle the FPR components.
20625 (rs6000_emit_prologue_components): Handle the FPR components.
20626 (rs6000_emit_epilogue_components): Handle the FPR components.
20627 (rs6000_set_handled_components): Handle the FPR components.
20628 (rs6000_emit_prologue): Don't output prologue code for those FPRs
20629 that are already separately shrink-wrapped.
20630 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
20631 that are already separately shrink-wrapped.
20633 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20636 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
20639 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
20640 (define_peephole2 for Altivec d-form load): Add peepholes to catch
20641 cases where the register allocator uses a move and an offsettable
20642 memory operation to/from a FPR register on ISA 2.06/2.07.
20643 (define_peephole2 for Altivec d-form store): Likewise.
20645 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
20648 * config/i386/mmx.md (*mov<mode>_internal): Enable
20649 alternatives 11, 12, 13 and 14 also for 32bit targets.
20650 Remove alternatives 15, 16, 17 and 18.
20651 * config/i386/sse.md (vec_concatv2di): Change
20652 alternative (!x, *y) to (x, ?!*Yn).
20654 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
20656 * dumpfile.h (enum dump_kind): Remove stray comma.
20658 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20660 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
20661 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
20662 predicate::num_conditions
20663 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
20664 (CHANGED): turn into predicate::changed.
20665 (agg_position_info): Move to ipa-predicate.h
20666 (add_condition, predicate::add_clause, predicate::operator &=,
20667 predicate::or_with, predicate::evaluate, predicate::probability,
20668 dump_condition, dump_clause, predicate::dump,
20669 predicate::remap_after_duplication, predicate::remap_after_inlining,
20670 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
20671 (evaluate_conditions_for_known_args): Update.
20672 (set_cond_stmt_execution_predicate): Update.
20673 * ipa-inline.h: Include ipa-predicate.h
20674 (condition, inline_param_summary, conditions, agg_position_info,
20675 predicate): Move to ipa-predicate.h
20676 * ipa-predicate.c: New file.
20677 * ipa-predicate.h: New file.
20679 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
20681 * final.c (leaf_function_p): Check we are not in a sequence.
20683 2017-05-18 Martin Liska <mliska@suse.cz>
20685 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
20686 * dumpfile.c (dump_register): Use new enum dump_kind.
20687 (get_dump_file_name): Likewise.
20688 (dump_enable_all): Likewise.
20689 (dump_switch_p_1): Likewise.
20690 (enable_rtl_dump_file): Remove usage of TDF_RTL.
20691 * dumpfile.h (enum dump_kind): New enum type.
20692 (struct dump_file_info): Create constructor and
20693 format fields and comments.
20694 * passes.c (pass_manager::register_one_dump_file):
20696 * statistics.c (statistics_early_init): Likewise.
20697 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
20698 TDF_TREE with TDF_SLIM.
20699 (gather_memory_references_ref): Likewise.
20701 2017-05-18 Martin Liska <mliska@suse.cz>
20703 * vec.h (struct vnull): Use it.
20705 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20707 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
20708 (true_predicate, false_predicate, true_predicate_p,
20709 false_predicate_p): Remove.
20710 (single_cond_predicate, not_inlined_predicate): Turn to member function
20712 (add_condition): Update.
20713 (add_clause): Turn to...
20714 (predicate::add_clause): ... this one; update; allow passing NULL
20716 (and_predicates): Turn to ...
20717 (predicate::operator &=): ... this one.
20718 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
20719 (or_predicates): Turn to ...
20720 (predicate::or_with): ... this one.
20721 (evaluate_predicate): Turn to ...
20722 (predicate::evaluate): ... this one.
20723 (predicate_probability): Turn to ...
20724 (predicate::probability): ... this one.
20725 (dump_condition): Update.
20726 (dump_predicate): Turn to ...
20727 (predicate::dump): ... this one.
20728 (account_size_time): Update.
20729 (edge_set_predicate): Update.
20730 (set_hint_predicate): UPdate.
20731 (evaluate_conditions_for_known_args): Update.
20732 (evaluate_properties_for_edge): Update.
20733 (remap_predicate_after_duplication): Turn to...
20734 (predicate::remap_after_duplication): ... this one.
20735 (remap_hint_predicate_after_duplication): Update.
20736 (inline_summary_t::duplicate): UPdate.
20737 (dump_inline_edge_summary): Update.
20738 (dump_inline_summary): Update.
20739 (set_cond_stmt_execution_predicate): Update.
20740 (set_switch_stmt_execution_predicate): Update.
20741 (compute_bb_predicates): Update.
20742 (will_be_nonconstant_expr_predicate): Update.
20743 (will_be_nonconstant_predicate): Update.
20744 (phi_result_unknown_predicate): Update.
20745 (predicate_for_phi_result): Update.
20746 (array_index_predicate): Update.
20747 (estimate_function_body_sizes): Update.
20748 (estimate_node_size_and_time): Update.
20749 (estimate_ipcp_clone_size_and_time): Update.
20750 (remap_predicate): Rename to ...
20751 (predicate::remap_after_inlining): ... this one.
20752 (remap_hint_predicate): Update.
20753 (inline_merge_summary): Update.
20754 (inline_update_overall_summary): Update.
20755 (estimate_size_after_inlining): Update.
20756 (read_predicate): Rename to ...
20757 (predicate::stream_in): ... this one.
20758 (read_inline_edge_summary): Update.
20759 (write_predicate): Rename to ...
20760 (predicate::stream_out): ... this one.
20761 (write_inline_edge_summary): Update.
20762 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
20763 (clause_t): Turn to uint32_t
20764 (predicate): Turn to class; implement constructor and operators
20766 (size_time_entry): Update.
20767 (inline_summary): Update.
20768 (inline_edge_summary): Update.
20770 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
20772 * fold-const.c (fold_binary_loc): Move transformation...
20773 * match.pd (C - X CMP X): ... here.
20775 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
20777 * config/sparc/sparc.c (sparc_option_override): Set function
20778 alignment for -mcpu=niagara7 to 64 to match the I$ line.
20779 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
20781 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
20783 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
20785 2017-05-18 Marek Polacek <polacek@redhat.com>
20788 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
20789 (pass_ubsan::execute): Call gimple_assign_single_p instead of
20790 gimple_assign_load_p.
20792 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
20794 PR middle-end/80692
20795 * real.c (do_compare): Give decimal_do_compare preference over
20796 comparing just the signs.
20798 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
20800 * doc/md.texi (Canonicalization of Instructions): Describe the
20801 canonical form of instructions that inherently set a condition
20804 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
20806 PR middle-end/80775
20807 * tree-cfg.c: Move deletion of unreachable case statements to after
20808 the merging of consecutive case labels.
20810 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
20812 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
20813 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
20814 restoring of callee-saved registers.
20816 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
20818 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
20819 * config/visium/visium.c (single_set_and_flags): Likewise.
20820 * config/visium/visium.md (Substitutions): Likewise.
20822 2017-05-17 Martin Liska <mliska@suse.cz>
20824 * cfg.c: Introduce dump_flags_t type and
20825 use it instead of int type.
20827 * cfghooks.c: Likewise.
20828 * cfghooks.h (struct cfg_hooks): Likewise.
20829 * cfgrtl.c: Likewise.
20830 * cfgrtl.h: Likewise.
20831 * cgraph.c (cgraph_node::get_body): Likewise.
20832 * coretypes.h: Likewise.
20833 * domwalk.c: Likewise.
20834 * domwalk.h: Likewise.
20835 * dumpfile.c (struct dump_option_value_info): Likewise.
20836 (dump_enable_all): Likewise.
20837 (dump_switch_p_1): Likewise.
20838 (opt_info_switch_p): Likewise.
20839 * dumpfile.h (enum tree_dump_index): Likewise.
20840 (struct dump_file_info): Likewise.
20841 * genemit.c: Likewise.
20842 * generic-match-head.c: Likewise.
20843 * gengtype.c (open_base_files): Likewise.
20844 * gimple-pretty-print.c: Likewise.
20845 * gimple-pretty-print.h: Likewise.
20846 * graph.c (print_graph_cfg): Likewise.
20847 * graphite-scop-detection.c (dot_all_sese): Likewise.
20848 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
20849 * loop-unroll.c (report_unroll): Likewise.
20850 * passes.c (pass_manager::register_one_dump_file): Likewise.
20851 * print-tree.c: Likewise.
20852 * statistics.c: Likewise.
20853 * tree-cfg.c: Likewise.
20854 * tree-cfg.h: Likewise.
20855 * tree-dfa.c: Likewise.
20856 * tree-dfa.h: Likewise.
20857 * tree-dump.c (dump_function): Likewise.
20858 * tree-dump.h (struct dump_info): Likewise.
20859 * tree-pretty-print.c: Likewise.
20860 * tree-pretty-print.h: Likewise.
20861 * tree-ssa-live.c: Likewise.
20862 * tree-ssa-live.h: Likewise.
20863 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
20864 * tree-vect-loop.c: Likewise.
20865 * tree-vect-slp.c: Likewise.
20867 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
20868 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20870 PR tree-optimization/80457
20871 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
20872 of all arguments to a statement as scalar_to_vec operations.
20873 (vectorizable_call): Adjust call to vect_model_simple_cost for
20875 (vectorizable_conversion): Likewise.
20876 (vectorizable_assignment): Likewise.
20877 (vectorizable_shift): Likewise.
20878 (vectorizable_operation): Likewise.
20879 (vectorizable_comparison): Likewise.
20880 (vect_is_simple_cond): Record the def types for operands.
20881 (vectorizable_condition): Likewise, call vect_model_simple_cost.
20882 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
20883 for statement argument count.
20885 2017-05-16 Carl Love <cel@us.ibm.com>
20887 * config/rs6000/rs6000-c: Add support for built-in functions
20888 vector unsigned long long vec_bperm (vector unsigned long long,
20889 vector unsigned char)
20890 vector signed long long vec_mule (vector signed int,
20892 vector unsigned long long vec_mule (vector unsigned int,
20893 vector unsigned int)
20894 vector signed long long vec_mulo (vector signed int,
20896 vector unsigned long long vec_mulo (vector unsigned int,
20897 vector unsigned int)
20898 vector signed char vec_sldw (vector signed char,
20899 vector signed char,
20901 vector unsigned char vec_sldw (vector unsigned char,
20902 vector unsigned char,
20904 vector signed short vec_sldw (vector signed short,
20905 vector signed short,
20907 vector unsigned short vec_sldw (vector unsigned short,
20908 vector unsigned short,
20910 vector signed int vec_sldw (vector signed int,
20913 vector unsigned int vec_sldw (vector unsigned int,
20914 vector unsigned int,
20916 vector signed long long vec_sldw (vector signed long long,
20917 vector signed long long,
20919 vector unsigned long long vec_sldw (vector unsigned long long,
20920 vector unsigned long long,
20922 * config/rs6000/rs6000-c: Add support for built-in functions
20923 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
20924 * config/rs6000/altivec.h: Add defintion for vec_sldw.
20925 * doc/extend.texi: Update the built-in documentation for the
20926 new built-in functions.
20928 2017-05-16 Marek Polacek <polacek@redhat.com>
20932 * tree.c (save_expr): Don't fold the expression.
20934 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
20936 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
20937 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
20938 and (?*y,m). Update insn attributes.
20940 2017-05-16 Martin Liska <mliska@suse.cz>
20942 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
20943 flags argument of print_gimple_stmt, print_gimple_expr,
20944 print_generic_stmt and print_generic_expr.
20945 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
20946 * coretypes.h: Likewise.
20947 * except.c (dump_eh_tree): Likewise.
20948 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
20949 * gimple-pretty-print.h: Likewise.
20950 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
20951 (backprop::push_to_worklist): Likewise.
20952 (backprop::pop_from_worklist): Likewise.
20953 (backprop::process_use): Likewise.
20954 (backprop::intersect_uses): Likewise.
20955 (note_replacement): Likewise.
20956 * gimple-ssa-store-merging.c
20957 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
20958 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
20959 (pass_store_merging::execute): Likewise.
20960 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
20961 (ssa_base_cand_dump_callback): Likewise.
20962 (dump_incr_vec): Likewise.
20963 (replace_refs): Likewise.
20964 (replace_mult_candidate): Likewise.
20965 (create_add_on_incoming_edge): Likewise.
20966 (create_phi_basis): Likewise.
20967 (insert_initializers): Likewise.
20968 (all_phi_incrs_profitable): Likewise.
20969 (introduce_cast_before_cand): Likewise.
20970 (replace_one_candidate): Likewise.
20971 * gimplify.c (gimplify_expr): Likewise.
20972 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
20973 (set_rename): Likewise.
20974 (rename_uses): Likewise.
20975 (copy_loop_phi_nodes): Likewise.
20976 (add_close_phis_to_merge_points): Likewise.
20977 (copy_loop_close_phi_args): Likewise.
20978 (copy_cond_phi_args): Likewise.
20979 (graphite_copy_stmts_from_block): Likewise.
20980 (translate_pending_phi_nodes): Likewise.
20981 * graphite-poly.c (print_pdr): Likewise.
20982 (dump_gbb_cases): Likewise.
20983 (dump_gbb_conditions): Likewise.
20984 (print_scop_params): Likewise.
20985 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
20986 (build_cross_bb_scalars_use): Likewise.
20987 (gather_bbs::before_dom_children): Likewise.
20988 * hsa-dump.c (dump_hsa_immed): Likewise.
20989 * ipa-cp.c (print_ipcp_constant_value): Likewise.
20990 (get_replacement_map): Likewise.
20991 * ipa-inline-analysis.c (dump_condition): Likewise.
20992 (estimate_function_body_sizes): Likewise.
20993 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
20994 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
20995 * ipa-prop.c (ipa_dump_param): Likewise.
20996 (ipa_print_node_jump_functions_for_edge): Likewise.
20997 (ipa_modify_call_arguments): Likewise.
20998 (ipa_modify_expr): Likewise.
20999 (ipa_dump_param_adjustments): Likewise.
21000 (ipa_dump_agg_replacement_values): Likewise.
21001 (ipcp_modif_dom_walker::before_dom_children): Likewise.
21002 * ipa-pure-const.c (check_stmt): Likewise.
21003 (pass_nothrow::execute): Likewise.
21004 * ipa-split.c (execute_split_functions): Likewise.
21005 * omp-offload.c (dump_oacc_loop_part): Likewise.
21006 (dump_oacc_loop): Likewise.
21007 * trans-mem.c (tm_log_emit): Likewise.
21008 (tm_memopt_accumulate_memops): Likewise.
21009 (dump_tm_memopt_set): Likewise.
21010 (dump_tm_memopt_transform): Likewise.
21011 * tree-cfg.c (gimple_verify_flow_info): Likewise.
21012 (print_loop): Likewise.
21013 * tree-chkp-opt.c (chkp_print_addr): Likewise.
21014 (chkp_gather_checks_info): Likewise.
21015 (chkp_get_check_result): Likewise.
21016 (chkp_remove_check_if_pass): Likewise.
21017 (chkp_use_outer_bounds_if_possible): Likewise.
21018 (chkp_reduce_bounds_lifetime): Likewise.
21019 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
21020 (chkp_mark_completed_bounds): Likewise.
21021 (chkp_register_incomplete_bounds): Likewise.
21022 (chkp_mark_invalid_bounds): Likewise.
21023 (chkp_maybe_copy_and_register_bounds): Likewise.
21024 (chkp_build_returned_bound): Likewise.
21025 (chkp_get_bound_for_parm): Likewise.
21026 (chkp_build_bndldx): Likewise.
21027 (chkp_get_bounds_by_definition): Likewise.
21028 (chkp_generate_extern_var_bounds): Likewise.
21029 (chkp_get_bounds_for_decl_addr): Likewise.
21030 * tree-chrec.c (chrec_apply): Likewise.
21031 * tree-data-ref.c (dump_data_reference): Likewise.
21032 (dump_subscript): Likewise.
21033 (dump_data_dependence_relation): Likewise.
21034 (analyze_overlapping_iterations): Likewise.
21035 * tree-inline.c (expand_call_inline): Likewise.
21036 (tree_function_versioning): Likewise.
21037 * tree-into-ssa.c (dump_defs_stack): Likewise.
21038 (dump_currdefs): Likewise.
21039 (dump_names_replaced_by): Likewise.
21040 (dump_update_ssa): Likewise.
21041 (update_ssa): Likewise.
21042 * tree-object-size.c (pass_object_sizes::execute): Likewise.
21043 * tree-parloops.c (build_new_reduction): Likewise.
21044 (try_create_reduction_list): Likewise.
21045 (ref_conflicts_with_region): Likewise.
21046 (oacc_entry_exit_ok_1): Likewise.
21047 (oacc_entry_exit_single_gang): Likewise.
21048 * tree-pretty-print.h: Likewise.
21049 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
21050 (get_scalar_evolution): Likewise.
21051 (add_to_evolution): Likewise.
21052 (get_loop_exit_condition): Likewise.
21053 (analyze_evolution_in_loop): Likewise.
21054 (analyze_initial_condition): Likewise.
21055 (analyze_scalar_evolution): Likewise.
21056 (instantiate_scev): Likewise.
21057 (number_of_latch_executions): Likewise.
21058 (gather_chrec_stats): Likewise.
21059 (final_value_replacement_loop): Likewise.
21060 (scev_const_prop): Likewise.
21061 * tree-sra.c (dump_access): Likewise.
21062 (disqualify_candidate): Likewise.
21063 (create_access): Likewise.
21064 (reject): Likewise.
21065 (maybe_add_sra_candidate): Likewise.
21066 (create_access_replacement): Likewise.
21067 (analyze_access_subtree): Likewise.
21068 (analyze_all_variable_accesses): Likewise.
21069 (sra_modify_assign): Likewise.
21070 (initialize_constant_pool_replacements): Likewise.
21071 (find_param_candidates): Likewise.
21072 (decide_one_param_reduction): Likewise.
21073 (replace_removed_params_ssa_names): Likewise.
21074 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
21075 * tree-ssa-copy.c (dump_copy_of): Likewise.
21076 (copy_prop_visit_cond_stmt): Likewise.
21077 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
21078 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
21079 (record_equivalences_from_stmt): Likewise.
21080 * tree-ssa-dse.c (compute_trims): Likewise.
21081 (delete_dead_call): Likewise.
21082 (delete_dead_assignment): Likewise.
21083 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
21084 (forward_propagate_into_cond): Likewise.
21085 (pass_forwprop::execute): Likewise.
21086 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
21087 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
21089 (move_computations_worker): Likewise.
21090 (execute_sm): Likewise.
21091 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
21092 (remove_exits_and_undefined_stmts): Likewise.
21093 (remove_redundant_iv_tests): Likewise.
21094 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
21095 (adjust_iv_update_pos): Likewise.
21096 * tree-ssa-math-opts.c (bswap_replace): Likewise.
21097 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
21098 (value_replacement): Likewise.
21099 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
21100 * tree-ssa-pre.c (print_pre_expr): Likewise.
21101 (get_representative_for): Likewise.
21102 (create_expression_by_pieces): Likewise.
21103 (insert_into_preds_of_block): Likewise.
21104 (eliminate_insert): Likewise.
21105 (eliminate_dom_walker::before_dom_children): Likewise.
21106 (eliminate): Likewise.
21107 (remove_dead_inserted_code): Likewise.
21108 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
21109 * tree-ssa-reassoc.c (get_rank): Likewise.
21110 (eliminate_duplicate_pair): Likewise.
21111 (eliminate_plus_minus_pair): Likewise.
21112 (eliminate_not_pairs): Likewise.
21113 (undistribute_ops_list): Likewise.
21114 (eliminate_redundant_comparison): Likewise.
21115 (update_range_test): Likewise.
21116 (optimize_range_tests_var_bound): Likewise.
21117 (optimize_vec_cond_expr): Likewise.
21118 (rewrite_expr_tree): Likewise.
21119 (rewrite_expr_tree_parallel): Likewise.
21120 (linearize_expr): Likewise.
21121 (break_up_subtract): Likewise.
21122 (linearize_expr_tree): Likewise.
21123 (attempt_builtin_powi): Likewise.
21124 (attempt_builtin_copysign): Likewise.
21125 (transform_stmt_to_copy): Likewise.
21126 (transform_stmt_to_multiply): Likewise.
21127 (dump_ops_vector): Likewise.
21128 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
21129 (print_scc): Likewise.
21130 (set_ssa_val_to): Likewise.
21131 (visit_reference_op_store): Likewise.
21132 (visit_use): Likewise.
21133 (sccvn_dom_walker::before_dom_children): Likewise.
21134 (run_scc_vn): Likewise.
21135 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
21137 (expr_hash_elt::print): Likewise.
21138 (const_and_copies::pop_to_marker): Likewise.
21139 (const_and_copies::record_const_or_copy_raw): Likewise.
21140 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
21141 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
21142 (dump_predicates): Likewise.
21143 (find_uninit_use): Likewise.
21144 (warn_uninitialized_phi): Likewise.
21145 (pass_late_warn_uninitialized::execute): Likewise.
21146 * tree-ssa.c (verify_vssa): Likewise.
21147 (verify_ssa): Likewise.
21148 (maybe_optimize_var): Likewise.
21149 * tree-vrp.c (dump_value_range): Likewise.
21150 (dump_all_value_ranges): Likewise.
21151 (dump_asserts_for): Likewise.
21152 (register_edge_assert_for_2): Likewise.
21153 (vrp_visit_cond_stmt): Likewise.
21154 (vrp_visit_switch_stmt): Likewise.
21155 (vrp_visit_stmt): Likewise.
21156 (vrp_visit_phi_node): Likewise.
21157 (simplify_cond_using_ranges_1): Likewise.
21158 (fold_predicate_in): Likewise.
21159 (evrp_dom_walker::before_dom_children): Likewise.
21160 (evrp_dom_walker::push_value_range): Likewise.
21161 (evrp_dom_walker::pop_value_range): Likewise.
21162 (execute_early_vrp): Likewise.
21164 2017-05-16 Richard Biener <rguenther@suse.de>
21166 * dwarf2out.c (loc_list_from_tree_1): Do not create
21167 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
21169 2017-05-16 Richard Biener <rguenther@suse.de>
21171 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
21173 (note_variable_value_in_expr): If we resolved the decl ref
21174 do not push to the stack.
21176 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
21178 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
21179 operations in fast-math mode.
21180 (vaddq_f16): Likewise.
21181 (vmul_f16): Likewise.
21182 (vmulq_f16): Likewise.
21183 (vsub_f16): Likewise.
21184 (vsubq_f16): Likewise.
21185 * config/arm/neon.md (add<mode>3): New.
21187 (fma:<VH:mode>3): New. Also remove outdated comment.
21190 2017-05-16 Martin Liska <mliska@suse.cz>
21194 * ipa-devirt.c (warn_types_mismatch): Fix typo.
21195 (odr_types_equivalent_p): Likewise.
21197 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
21199 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
21201 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
21204 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
21205 non-interunit SSE move alternatives with '?'.
21206 (zero-extendsidi peephole2): New peephole to skip intermediate
21207 general register in SSE zero-extend sequence.
21209 2017-05-15 Jeff Law <law@redhat.com>
21211 * reorg.c (relax_delay_slots): Create a new variable to hold
21212 the temporary target rather than clobbering TARGET_LABEL.
21214 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
21215 missing argument to extract_bit_field call.
21216 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
21218 2017-05-15 Martin Liska <mliska@suse.cz>
21221 * gcc.c (process_command): Do not allow empty argument of -o option.
21223 2017-05-15 Renlin Li <renlin.li@arm.com>
21225 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
21226 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
21227 * config/aarch64/constraints.md (Usf): Add long call check.
21228 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
21229 (call_value): Likewise.
21230 (sibcall): Likewise.
21231 (sibcall_value): Likewise.
21233 (call_value_insn): New.
21234 (sibcall_insn): Update rtx pattern.
21235 (sibcall_value_insn): Likewise.
21236 (call_internal): Remove.
21237 (call_value_internal): Likewise.
21238 (sibcall_internal): Likewise.
21239 (sibcall_value_internal): Likewise.
21240 (call_reg): Likewise.
21241 (call_symbol): Likewise.
21242 (call_value_reg): Likewise.
21243 (call_value_symbol): Likewise.
21245 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
21248 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
21250 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
21252 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
21253 compatible with CCGOCmode and with CCZmode.
21255 2017-05-14 Martin Sebor <msebor@redhat.com>
21257 PR middle-end/77671
21258 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
21259 (gimple_fold_builtin_snprintf): Same.
21260 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
21261 (gimple_fold_builtin_snprintf): Same.
21262 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
21263 of character types.
21264 (is_call_safe): New function.
21265 (try_substitute_return_value): Call it.
21266 (try_simplify_call): New function.
21267 (pass_sprintf_length::handle_gimple_call): Call it.
21269 2017-05-14 Martin Sebor <msebor@redhat.com>
21271 PR middle-end/80669
21272 * builtins.c (expand_builtin_stpncpy): Simplify.
21274 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
21276 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
21277 * config/i386/i386.h
21278 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
21279 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
21280 (struct machine_function): Add new members call_ms2sysv,
21281 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
21282 (struct machine_frame_state): New fields sp_realigned and
21283 sp_realigned_offset.
21284 * config/i386/i386.c
21285 (enum xlogue_stub): New enum.
21286 (enum xlogue_stub_sets): New enum.
21287 (class xlogue_layout): New class.
21288 (struct ix86_frame): New fields stack_realign_allocate_offset,
21289 stack_realign_offset and outlined_save_offset. Modify comments to
21290 detail stack layout when using out-of-line stubs.
21291 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
21292 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
21293 -mcall-ms2sysv-xlogues.
21294 (stub_managed_regs): New static variable.
21295 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
21296 registers managed by out-of-line stub.
21297 (disable_call_ms2sysv_xlogues): New function.
21298 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
21299 m->call_ms2sysv when appropriate and compute frame layout for
21301 (sp_valid_at, fp_valid_at): New inline functions.
21302 (choose_basereg): New function.
21303 (choose_baseaddr): Add align parameter, use choose_basereg and modify
21305 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
21306 Use align parameter of choose_baseaddr to generated aligned SSE movs
21308 (pro_epilogue_adjust_stack): Modify to track
21309 machine_frame_state::sp_realigned.
21310 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
21311 (ix86_nsaved_sseregs): Likewise.
21312 (ix86_emit_save_regs): Likewise.
21313 (ix86_emit_save_regs_using_mov): Likewise.
21314 (ix86_emit_save_sse_regs_using_mov): Likewise.
21315 (get_scratch_register_on_entry): Likewise.
21316 (gen_frame_set): New function.
21317 (gen_frame_load): Likewise.
21318 (gen_frame_store): Likewise.
21319 (emit_outlined_ms2sysv_save): Likewise.
21320 (emit_outlined_ms2sysv_restore): Likewise.
21321 (ix86_expand_prologue): Modify stack re-alignment code and call
21322 emit_outlined_ms2sysv_save when appropriate.
21323 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
21324 parameter rtx_insn *insn, which allows the function to be used to only
21325 generate the notes.
21326 (ix86_expand_epilogue): Modify validity checks of frame and stack
21327 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
21328 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
21329 * config/i386/predicates.md
21330 (save_multiple): New predicate.
21331 (restore_multiple): Likewise.
21332 * config/i386/sse.md
21333 (save_multiple<mode>): New pattern.
21334 (save_multiple_realign<mode>): Likewise.
21335 (restore_multiple<mode>): Likewise.
21336 (restore_multiple_and_return<mode>): Likewise.
21337 (restore_multiple_leave_return<mode>): Likewise.
21338 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
21340 2017-05-14 Julia Koval <julia.koval@intel.com>
21342 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
21343 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
21344 (__builtin_ia32_xsetbv): New builtins.
21345 * config/i386/i386.c (ix86_expand_special_args_builtin):
21347 (ix86_expand_builtin): Special expand for new intrinsics.
21348 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
21349 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
21350 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
21352 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21354 * cfganal.c (inverted_post_order_compute): Change argument type
21356 * cfganal.h (inverted_post_order_compute): Adjust prototype.
21357 * df-core.c (rest_of_handle_df_initialize): Adjust.
21358 (rest_of_handle_df_finish): Likewise.
21359 (df_analyze_1): Likewise.
21360 (df_analyze): Likewise.
21361 (loop_inverted_post_order_compute): Change argument to be a vec *.
21362 (df_analyze_loop): Adjust.
21363 (df_get_n_blocks): Likewise.
21364 (df_get_postorder): Likewise.
21365 * df.h (struct df_d): Change field to be a vec.
21366 * lcm.c (compute_laterin): Adjust.
21367 (compute_available): Likewise.
21368 * lra-lives.c (lra_create_live_ranges_1): Likewise.
21369 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
21370 * tree-ssa-pre.c (compute_antic): Likewise.
21372 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21374 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
21375 (depth_first_search::depth_first_search): Change structure init
21376 function to this constructor.
21377 (depth_first_search::add_bb): Rename function to this member.
21378 (depth_first_search::execute): Likewise.
21379 (flow_dfs_compute_reverse_finish): Adjust.
21381 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21383 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
21384 (longest_simple_path): Likewise.
21385 * shrink-wrap.c (spread_components): Likewise.
21386 (disqualify_problematic_components): Likewise.
21387 (emit_common_heads_for_components): Likewise.
21388 (emit_common_tails_for_components): Likewise.
21389 (insert_prologue_epilogue_for_components): Likewise.
21391 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21393 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
21396 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21398 * df-core.c (df_set_blocks): Start using auto_bitmap.
21399 (df_compact_blocks): Likewise.
21400 * df-problems.c (df_rd_confluence_n): Likewise.
21401 * df-scan.c (df_insn_rescan_all): Likewise.
21402 (df_process_deferred_rescans): Likewise.
21403 (df_update_entry_block_defs): Likewise.
21404 (df_update_exit_block_uses): Likewise.
21405 (df_entry_block_bitmap_verify): Likewise.
21406 (df_exit_block_bitmap_verify): Likewise.
21407 (df_scan_verify): Likewise.
21408 * lra-constraints.c (lra_constraints): Likewise.
21409 (undo_optional_reloads): Likewise.
21410 (lra_undo_inheritance): Likewise.
21411 * lra-remat.c (calculate_gen_cands): Likewise.
21412 (do_remat): Likewise.
21413 * lra-spills.c (assign_spill_hard_regs): Likewise.
21414 (spill_pseudos): Likewise.
21415 * tree-ssa-pre.c (bitmap_set_and): Likewise.
21416 (bitmap_set_subtract_values): Likewise.
21418 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21420 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
21421 management with auto_bitmap.
21422 (fix_inter_tick): Likewise.
21423 (fix_recovery_deps): Likewise.
21424 * ira.c (add_store_equivs): Likewise.
21425 (find_moveable_pseudos): Likewise.
21426 (split_live_ranges_for_shrink_wrap): Likewise.
21427 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
21428 (rtx_reuse_manager::seen_def_p): Likewise.
21429 (rtx_reuse_manager::set_seen_def): Likewise.
21430 * print-rtl.h (class rtx_reuse_manager): Likewise.
21432 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21434 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
21436 (migrate_btr_def): Likewise.
21437 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
21438 * df-core.c (loop_post_order_compute): Likewise.
21439 (loop_inverted_post_order_compute): Likewise.
21440 * hsa-common.h: Likewise.
21441 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
21442 * init-regs.c (initialize_uninitialized_regs): Likewise.
21443 * ipa-inline.c (resolve_noninline_speculation): Likewise.
21444 (inline_small_functions): Likewise.
21445 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
21446 * ira.c (combine_and_move_insns): Likewise.
21447 (build_insn_chain): Likewise.
21448 * loop-invariant.c (find_invariants): Likewise.
21449 * lower-subreg.c (propagate_pseudo_copies): Likewise.
21450 * predict.c (tree_predict_by_opcode): Likewise.
21451 (predict_paths_leading_to): Likewise.
21452 (predict_paths_leading_to_edge): Likewise.
21453 (estimate_loops_at_level): Likewise.
21454 (estimate_loops): Likewise.
21455 * shrink-wrap.c (try_shrink_wrapping): Likewise.
21456 (spread_components): Likewise.
21457 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
21458 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
21459 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
21460 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
21461 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
21462 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
21463 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
21464 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
21465 (mark_threaded_blocks): Likewise.
21466 (thread_through_all_blocks): Likewise.
21467 * tree-ssa.c (verify_ssa): Likewise.
21468 (execute_update_addresses_taken): Likewise.
21469 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
21471 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21473 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
21475 (post_order_compute): Likewise.
21476 (inverted_post_order_compute): Likewise.
21477 (pre_and_rev_post_order_compute_fn): Likewise.
21479 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21481 * genrecog.c (int_set::int_set): Explicitly construct our
21482 auto_vec base class.
21483 * vec.h (auto_vec::auto_vec): New constructor.
21485 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21487 * bitmap.h (class auto_bitmap): New constructor taking
21488 bitmap_obstack * argument.
21490 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21492 * bitmap.h (class auto_bitmap): Change type of m_bits to
21493 bitmap_head, and adjust ctor / dtor and member operators.
21495 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
21497 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
21498 when returned register mode doesn't match original mode.
21500 2017-05-12 Jeff Law <law@redhat.com>
21501 Jakub Jelinek <jakub@redhat.com>
21503 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
21504 we look for cc setter after the compare-elim changes.
21505 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
21506 within the vector to match what compare-elim now expects.
21507 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
21508 (xorsi3_flags, one_cmplsi2_flags): Likewise.
21510 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
21511 after the compare-elim changes.
21512 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
21513 the vector to match what compare-elim now expects.
21514 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
21515 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
21516 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
21517 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
21518 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
21520 * config/visium/visium.c (single_set_and_flags): Fix where
21521 we look for cc setter after the compare-elim changes.
21522 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
21523 with the vector to match what compare-elim now expects.
21524 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
21525 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
21526 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
21527 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
21528 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
21529 (neg<mode>2_insn_set_overflow): Likewise.
21531 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
21533 PR middle-end/79794
21534 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
21535 maybe_expand_insn call, set ops[0].target. If still set after call,
21536 set alt_rtl. Add extra arg to recursive calls.
21537 (extract_bit_field): Add alt_rtl argument. Pass to
21539 * expmed.h (extract_bit_field): Fix prototype.
21540 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
21541 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
21542 to extract_bit_field_calls.
21543 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
21544 Pass alt_rtl to extract_bit_field calls.
21545 * calls.c (store_unaligned_arguments_into_psuedos)
21546 load_register_parameters): Pass extra NULL to extract_bit_field calls.
21547 * optabs.c (maybe_legitimize_operand): Clear op->target when call
21549 * optabs.h (struct expand_operand): Add target bitfield.
21551 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21553 * compare-elim.c (try_eliminate_compare): Canonicalize
21554 operation with embedded compare to
21555 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
21556 (set (reg) (operation)].
21558 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
21560 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21563 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
21564 cost of adding a carry flag for ADC instruction.
21565 [case MINUS]: Ignore the cost of subtracting a carry flag
21566 for SBB instruction.
21568 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
21570 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
21572 * config/rs6000/bmiintrin.h: New file.
21573 * config/rs6000/bmi2intrin.h: New file.
21574 * config/rs6000/x86intrin.h: New file.
21576 2017-05-12 Jeff Law <law@redhat.com>
21578 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
21581 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
21583 PR middle-end/80707
21584 * tree-cfg.c: Remove cfg edges of unreachable case statements.
21586 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21588 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21589 early expansion of vector divide builtins.
21590 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
21591 builtins identified as having unsigned arguments.
21593 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21595 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
21596 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
21597 expansion of vector logical operations (and, andc, or, xor,
21600 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21602 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
21603 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
21605 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21607 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21608 early GIMPLE expansion of vector multiplies.
21610 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21612 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
21613 TARGET_HAVE_MOVT conditional.
21614 (movt splitter): Likewise.
21616 2017-05-12 Richard Biener <rguenther@suse.de>
21618 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
21619 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21620 Fold all stmts not inplace.
21622 2017-05-12 Richard Biener <rguenther@suse.de>
21624 PR tree-optimization/80713
21625 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
21626 inserted_exprs bit for not removed stmts.
21628 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
21630 PR middle-end/69921
21631 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
21632 parallelized" attribute for parallelized OpenACC kernels.
21633 * omp-offload.c (execute_oacc_device_lower): Use it.
21635 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
21636 Set "oacc kernels" attribute.
21637 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
21638 parameter. Adjust all users.
21639 (oacc_fn_attrib_kernels_p): Remove function.
21640 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
21641 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
21642 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
21643 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
21644 assert "oacc kernels" attribute is set.
21646 2017-05-11 Carl Love <cel@us.ibm.com>
21648 * config/rs6000/rs6000-c: Add support for built-in functions
21649 vector unsigned char vec_popcnt (vector signed char)
21650 vector unsigned char vec_popcnt (vector unsigned char)
21651 vector unsigned short vec_popcnt (vector signed short)
21652 vector unsigned short vec_popcnt (vector unsigned short)
21653 vector unsigned int vec_popcnt (vector signed int)
21654 vector unsigned int vec_popcnt (vector unsigned int)
21655 vector unsigned long long vec_popcnt (vector signed long long)
21656 vector unsigned long long vec_popcnt (vector unsigned long long)
21657 vector signed long long vec_slo (vector signed long long,
21658 vector signed char)
21659 vector signed long long vec_slo (vector signed long long,
21660 vector unsigned char)
21661 vector unsigned long long vec_slo (vector unsigned long long,
21662 vector signed char)
21663 vector unsigned long long vec_slo (vector unsigned long long,
21664 vector unsigned char)
21665 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
21666 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
21667 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
21668 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
21669 * doc/extend.texi: Update the built-in documentation file for the
21670 new built-in functions.
21672 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
21674 * attribs.h (sorted_attr_string): Move machine independent
21675 functions for target clone support from the i386 port to common
21676 code. Rename ix86_function_versions to common_function_versions.
21677 Rename make_name to make_unique_name.
21678 (common_function_versions): Likewise.
21679 (make_unique_name): Likewise.
21680 (make_dispatcher_decl): Likewise.
21681 (is_function_default_version): Likewise.
21682 * attribs.c (attr_strcmp): Likewise.
21683 (sorted_attr_string): Likewise.
21684 (common_function_versions): Likewise.
21685 (make_unique_name): Likewise.
21686 (make_dispatcher_decl): Likewise.
21687 (is_function_default_version): Likewise.
21688 * config/i386/i386.c (attr_strcmp): Likewise.
21689 (sorted_attr_string): Likewise.
21690 (ix86_function_versions): Likewise.
21691 (make_name): Likewise.
21692 (make_dispatcher_decl): Likewise.
21693 (is_function_default_version): Likewise.
21694 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
21696 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21699 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21700 Account for direct move costs for vec_construct of integer
21703 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
21706 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
21707 (UNSPEC_STX_ATOMIC): Ditto.
21708 (loaddi_via_sse): New insn.
21709 (storedi_via_sse): Ditto.
21710 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
21711 Update corresponding peephole2 patterns.
21712 (atomic_storedi_fpu): Ditto.
21714 2017-05-11 Julia Koval <julia.koval@intel.com>
21716 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
21717 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
21719 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
21720 (__builtin_ia32_rsqrt14ss_mask): New builtins.
21721 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
21723 2017-05-11 Nathan Sidwell <nathan@acm.org>
21725 * graphite-poly.c: Include dumpfile.h.
21727 * dumpfle.h (dump_function): Declare here ...
21728 * tree-dump.h (dump_function): ... not here.
21729 * dumpfile.c: #include tree-cfg.h.
21730 (dump_function): Move here from ...
21731 * tree-dump.c (dump_function): ... here.
21732 * gimplify.c: #include splay-tree.h, not tree-dump.h.
21733 * graphite-poly.c: Don't include tree-dump.h.
21734 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
21735 * print-tree.c: Likewise.
21736 * stor-layout.c: Likewise.
21737 * tree-nested.c: Likewise.
21739 * dumpfile.c (dump_start): Use TDF_FLAGS.
21740 (dump_enable_all): Fix TDF_KIND check thinko.
21742 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
21744 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21745 array entries to represent two legal parameterizations of the
21746 overloaded __builtin_cmpb function, as represented by the
21747 P6_OV_BUILTIN_CMPB constant.
21748 (altivec_resolve_overloaded_builtin): Add special case handling
21749 for the __builtin_cmpb function, as represented by the
21750 P6_OV_BUILTIN_CMPB constant.
21751 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
21752 (BU_P6_64BIT_2): New macro.
21753 (BU_P6_OVERLOAD_2): New macro
21754 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
21755 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
21756 (CMPB): Add overload support to represent both 32-bit and 64-bit
21757 compare-bytes function.
21758 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
21759 support for TARGET_CMPB.
21760 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
21761 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21762 documentation of the __builtin_cmpb overloaded built-in function.
21764 2017-05-11 Richard Biener <rguenther@suse.de>
21766 PR tree-optimization/80705
21767 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
21768 bases are not vectorizable.
21770 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21772 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
21773 when counting register pressure.
21775 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21777 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
21778 (iv_ca_more_deps): Renamed to ...
21779 (iv_ca_compare_deps): ... this.
21780 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
21782 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21784 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
21786 (determine_group_iv_costs): ... here.
21787 (find_inv_vars_cb): Record inv var if it's not recorded before.
21789 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21791 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
21792 (get_shiftadd_cost): Ditto.
21794 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21796 * tree-ssa-address.c: Include header file.
21797 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
21799 (add_to_parts): Refactor.
21800 (addr_to_parts): New parameter. Update use of move_hint_to_base.
21801 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
21804 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21806 PR tree-optimization/53090
21807 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
21809 (extract_cond_operands): Detect condition with IV on both sides
21810 and return COMP_IV_EXPR_2.
21811 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
21812 (rewrite_use_compare): Simplify by removing call to function
21813 extract_cond_operands.
21815 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21817 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
21818 (extract_cond_operands): Detect condition comparing against non-
21819 invariant bound and return appropriate enum value.
21820 (find_interesting_uses_cond): Update use of extract_cond_operands.
21821 Handle its return value accordingly.
21822 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
21824 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21826 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
21827 nonlinear iv_use computation in loop invariant sensitive way.
21829 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21831 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
21832 (find_iv_candidates): Call relate_compare_use_with_all_cands.
21834 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21836 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
21837 (dump_cand): Support iv_cand.inv_exprs.
21838 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
21840 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
21843 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21845 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
21847 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
21848 as local function. Include necessary header files.
21849 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
21851 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21853 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
21855 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21857 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
21858 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
21859 RSHIFT_EXPR and BIT_NOT_EXPR.
21861 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21863 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
21864 (adjust_setup_cost): New parameter supporting round up adjustment.
21865 (struct address_cost_data): Delete.
21866 (force_expr_to_var_cost): Don't bound cost with spill_cost.
21867 (split_address_cost, ptr_difference_cost): Delete.
21868 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
21869 (struct ainc_cost_data): New struct.
21870 (get_address_cost_ainc): New function.
21871 (get_address_cost, get_computation_cost): Reimplement.
21872 (determine_group_iv_cost_address): Record inv_expr for all uses of
21874 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
21875 (iv_ca_has_deps): Reimplemented to ...
21876 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
21878 (iv_ca_extend): Call iv_ca_more_deps.
21880 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21882 * tree-ssa-address.c (struct mem_address): Move to header file.
21883 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
21884 * tree-ssa-address.h (struct mem_address): Move from C file.
21885 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
21887 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21889 * tree-affine.h (aff_combination_type): New interface.
21890 (aff_combination_zero_p): Remove static.
21891 (aff_combination_const_p): New interface.
21892 (aff_combination_singleton_var_p): New interfaces.
21894 2017-05-11 Richard Biener <rguenther@suse.de>
21896 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21897 Skip unreachable blocks and destinations.
21898 (eliminate): Move stmt removal and fixup ...
21899 (fini_eliminate): ... here. Skip inserted exprs.
21900 (pass_pre::execute): Move fini_pre after fini_eliminate.
21901 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
21902 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
21903 PRE to get rid of dead code that has invalid SSA form and
21904 split critical edges again.
21906 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21908 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21910 2017-05-11 Richard Biener <rguenther@suse.de>
21912 * passes.c (execute_function_todo): Verify loops if they are
21913 said to be up-to-date.
21914 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
21915 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
21917 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
21920 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
21921 handle calling assemble_external ourself.
21924 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
21925 modes with zero size. Enhance comment.
21927 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21929 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
21930 built-ins for vec_xl and vec_xst with short and char pointer
21933 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
21935 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
21936 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
21937 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
21938 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
21939 (_mm_maskz_min_round_ss): New intrinsics.
21940 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
21941 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
21942 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
21943 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
21944 (__builtin_ia32_minss_mask_round): New builtins.
21945 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21946 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21947 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
21949 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
21950 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
21952 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
21955 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
21957 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
21958 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
21959 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
21960 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
21961 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
21962 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
21963 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
21964 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21965 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21966 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
21967 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
21968 (__builtin_ia32_mulss_mask_round): New builtins.
21969 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21970 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21971 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
21973 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
21974 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21976 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21979 2017-05-10 Julia Koval <julia.koval@intel.com>
21981 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
21982 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
21983 (_mm256_setr_m128i): New intrinsics.
21985 2017-05-10 Julia Koval <julia.koval@intel.com>
21987 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
21988 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
21989 (_mm_maskz_rcp14_ss): New intrinsics.
21990 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
21991 (__builtin_ia32_rcp14ss_mask): New builtins.
21992 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
21994 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
21996 PR tree-optimization/51513
21997 * tree-cfg.c (gimple_seq_unreachable_p): New function.
21998 (assert_unreachable_fallthru_edge_p): Use it.
21999 (group_case_labels_stmt): Likewise.
22000 * tree-cfg.h: Prototype it.
22001 * stmt.c: Include cfghooks.h and tree-cfg.h.
22002 (emit_case_dispatch_table) <gap_label>: New local variable.
22003 Use it to fill dispatch table gaps.
22004 Test for default_label before updating probabilities.
22005 (expand_case) <default_label>: Remove unneeded initialization.
22006 Test for unreachable default case statement and remove its edge.
22007 Set default_label accordingly.
22008 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
22010 2017-05-10 Carl Love <cel@us.ibm.com>
22012 * config/rs6000/rs6000-c: Add support for built-in functions
22013 vector signed char vec_neg (vector signed char)
22014 vector signed short int vec_neg (vector short int)
22015 vector signed int vec_neg (vector signed int)
22016 vector signed long long vec_neg (vector signed long long)
22017 vector float vec_neg (vector float)
22018 vector double vec_neg (vector double)
22019 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
22021 * config/rs6000/altivec.h: Add define for vec_neg
22022 * doc/extend.texi: Update the built-in documentation for the
22023 new built-in functions.
22025 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22027 PR tree-optimization/77644
22028 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
22030 2017-05-10 Nathan Sidwell <nathan@acm.org>
22032 * dumpfile.h (TDI_lang_all): New.
22033 (TDF_KIND): New. Renumber others
22034 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
22036 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
22038 (get_dump_file_name): Adjust suffix generation.
22039 (dump_enable_all): Use TDF_KIND.
22040 * doc/invoke.texi (-fdump-lang-all): Document.
22042 * dumpfile.h: Tabify.
22044 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
22047 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
22048 Move member access before delete.
22050 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
22052 * tree-inline.c (expand_call_inline): Split block at stmt
22055 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
22058 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
22059 are now unused after splitting mov{sf,sd}_hardfloat.
22060 (f32_lr2): Likewise.
22061 (f32_lm): Likewise.
22062 (f32_lm2): Likewise.
22063 (f32_li): Likewise.
22064 (f32_li2): Likewise.
22065 (f32_lv): Likewise.
22066 (f32_sr): Likewise.
22067 (f32_sr2): Likewise.
22068 (f32_sm): Likewise.
22069 (f32_sm2): Likewise.
22070 (f32_si): Likewise.
22071 (f32_si2): Likewise.
22072 (f32_sv): Likewise.
22073 (f32_dm): Likewise.
22074 (f32_vsx): Likewise.
22075 (f32_av): Likewise.
22076 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
22077 For movsf, order stores so the VSX stores occur before the GPR
22078 store which encourages the register allocator to use a traditional
22079 FPR instead of a GPR. For movsd, order the stores so that the GPR
22080 store comes before the VSX stores to allow the power6 to work.
22081 This is due to the power6 not having a 32-bit integer store
22082 instruction from a FPR.
22083 (movsf_hardfloat): Likewise.
22084 (movsd_hardfloat): Likewise.
22086 2017-05-09 Martin Sebor <msebor@redhat.com>
22088 PR translation/80280
22089 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
22092 PR translation/80280
22093 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
22094 data member added in r247778.
22095 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
22097 2017-05-09 Nathan Sidwell <nathan@acm.org>
22099 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
22101 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
22104 2017-05-09 Marek Polacek <polacek@redhat.com>
22106 * doc/invoke.texi: Fix typo.
22108 2017-05-09 Richard Biener <rguenther@suse.de>
22110 * tree-vrp.c (vrp_val_is_max): Adjust comment.
22111 (vrp_val_is_min): Likewise.
22112 (set_value_range_to_value): Likewise.
22113 (set_value_range_to_nonnegative): Likewise.
22114 (gimple_assign_nonzero_p): Likewise.
22115 (gimple_stmt_nonzero_p): Likewise.
22116 (vrp_int_const_binop): Likewise. Remove unreachable case.
22117 (adjust_range_with_scev): Adjust comments.
22118 (compare_range_with_value): Likewise.
22119 (extract_range_from_phi_node): Likewise.
22120 (test_for_singularity): Likewise.
22122 2017-05-09 Richard Biener <rguenther@suse.de>
22124 * tree-vrp.c (get_single_symbol): Add assert that we don't
22125 get overflowed constants as invariant part.
22126 (compare_values_warnv): Add comment before the TREE_NO_WARNING
22127 checks. Use wi::cmp instead of recursing for integer constants.
22128 (compare_values): Just ignore whether we assumed undefined
22129 overflow instead of failing the compare.
22130 (extract_range_for_var_from_comparison_expr): Add comment before the
22131 TREE_NO_WARNING sets.
22132 (test_for_singularity): Likewise.
22133 (extract_range_from_comparison): Do not disable optimization
22134 when we assumed undefined overflow.
22135 (extract_range_basic): Remove init of unused var.
22137 2017-05-09 Richard Biener <rguenther@suse.de>
22139 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
22140 (extract_range_from_multiplicative_op_1): Adjust.
22141 (extract_range_from_binary_expr_1): Use int_const_binop.
22143 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
22146 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
22147 rs6000_store_data_bypass_p in seven define_bypass directives and
22148 in several comments.
22149 * config/rs6000/rs6000-protos.h: Add prototype for
22150 rs6000_store_data_bypass_p function.
22151 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
22152 function implements slightly different (rs6000-specific) semantics
22153 than store_data_bypass_p, returning false rather than aborting
22154 with assertion error when arguments do not satisfy the
22155 requirements of store data bypass.
22156 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
22157 rs6000_store_data_bypass_p.
22159 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
22161 * config/xtensa/xtensa-protos.h
22162 (xtensa_initial_elimination_offset): New declaration.
22163 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22164 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
22165 macro definition, add case for FRAME_POINTER_REGNUM when
22166 FRAME_GROWS_DOWNWARD.
22167 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
22168 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
22169 xtensa_initial_elimination_offset.
22171 2017-05-08 Nathan Sidwell <nathan@acm.org>
22173 * doc/invoke.texi: Alphabetize -fdump options.
22175 2017-05-08 Martin Sebor <msebor@redhat.com>
22177 PR translation/80280
22178 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
22180 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
22182 * target.def (compute_frame_layout): New optional target hook.
22183 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
22184 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
22185 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
22187 * reload1.c (verify_initial_elim_offsets): Likewise.
22188 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
22189 (use_simple_return_p): Call arm_compute_frame_layout if needed.
22190 (arm_get_frame_offsets): Split up into this ...
22191 (arm_compute_frame_layout): ... and this function.
22193 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
22195 * config/aarch64/constraints.md (Usa): New constraint.
22196 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
22198 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
22200 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
22201 with_multilib_list after it has been checked.
22203 2017-05-08 Richard Biener <rguenther@suse.de>
22205 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
22206 (bitmap_set_subtract_values): Likewise.
22208 2017-05-08 Richard Biener <rguenther@suse.de>
22210 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
22211 (gimple_assign_nonzero): ... this and remove strict_overflow_p
22213 (gimple_stmt_nonzero_warnv_p): Rename to ...
22214 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
22216 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
22217 (extract_range_basic): Adjust, do not disable propagation on
22218 strict overflow sensitive simplification.
22219 (vrp_visit_cond_stmt): Likewise.
22221 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
22223 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
22224 body size unconditionally.
22226 2017-05-07 Jeff Law <law@redhat.com>
22229 2017-05-06 Jeff Law <law@redhat.com>
22230 PR tree-optimization/78496
22231 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
22234 PR tree-optimization/78496
22235 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
22236 (simplify_stmt_using_ranges): Call it.
22237 (vrp_dom_walker::before_dom_children): Extract equivalences
22238 from an ASSERT_EXPR with an equality comparison against a
22241 2017-05-06 Jeff Law <law@redhat.com>
22243 PR tree-optimization/78496
22244 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
22247 PR tree-optimization/78496
22248 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
22249 (simplify_stmt_using_ranges): Call it.
22250 (vrp_dom_walker::before_dom_children): Extract equivalences
22251 from an ASSERT_EXPR with an equality comparison against a
22254 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
22256 * lra-constraints.c (lra_copy_reg_equiv): New function.
22257 (split_reg): Use it to copy equivalence information from the
22258 original register to the spill register.
22260 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
22262 PR rtl-optimization/75964
22263 * simplify-rtx.c (simplify_const_relational_operation): Remove
22264 invalid handling of comparisons of integer ABS.
22266 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
22268 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
22269 initialize to zero.
22270 (init_regs): Remove declaration.
22271 (function_arg_advance_32): Initialize error_p as boolean variable.
22273 2017-05-05 Nathan Sidwell <nathan@acm.org>
22275 * store-motion.c (remove_reachable_equiv_notes): Reformat long
22276 lines. Use for (;;).
22278 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22280 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
22281 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
22282 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
22283 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
22284 VF=2 that require versioning.
22286 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22288 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
22291 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22293 * diagnostic.h (diagnostic_override_option_index): Convert from
22294 macro to inline function.
22296 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22298 * diagnostic.c (last_module_changed_p): New function.
22299 (set_last_module): New function.
22300 (diagnostic_report_current_module): Convert macro usage to
22301 the above functions.
22302 * diagnostic.h (diagnostic_context::last_module): Strengthen
22303 from const line_map * to const line_map_ordinary *.
22304 (diagnostic_last_module_changed): Delete macro.
22305 (diagnostic_set_last_module): Delete macro.
22307 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22309 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
22310 with diagnostic_report_diagnostic.
22311 (diagnostic_n_impl_richloc): Likewise.
22312 * diagnostic.h (report_diagnostic): Delete macro.
22313 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
22314 with diagnostic_report_diagnostic.
22315 * substring-locations.c (format_warning_va): Likewise.
22317 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22319 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
22320 save/restor of format_spec. Move option-printing code to...
22321 (print_option_information): ...this new function, and
22322 reimplement by simply printing to the pretty_printer,
22323 rather than appending to the format string.
22325 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22327 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
22328 handling logic into...
22329 (update_effective_level_from_pragmas): ...this new function.
22331 2017-05-04 Andrew Waterman <andrew@sifive.com>
22333 * config/riscv/riscv.opt (mstrict-align): New option.
22334 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
22335 (SLOW_UNALIGNED_ACCESS): Define.
22336 (riscv_slow_unaligned_access): Declare.
22337 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
22339 (riscv_slow_unaligned_access): New variable.
22340 (rocket_tune_info): Set slow_unaligned_access to true.
22341 (optimize_size_tune_info): Set slow_unaligned_access to false.
22342 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
22343 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
22344 (riscv_option_override): Set riscv_slow_unaligned_access.
22345 * doc/invoke.texi: Add -mstrict-align to RISC-V.
22347 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
22349 * config/riscv/riscv.md: Unify indentation.
22351 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22356 * config/rs6000/rs6000.md (u code attribute): Add FIX and
22358 (extendsi<mode>2): Add support for doing sign extension via
22359 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
22360 don't have ISA 3.0 instructions.
22361 (extendsi<mode>2 splitter): Likewise.
22362 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
22363 generate the normal insns since SImode can now go in vector
22364 registers. Disallow the special UNSPECs needed for previous
22365 machines to hide SImode being used. Add new insns
22366 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
22367 (fix_trunc<mode>si2_stfiwx): Likewise.
22368 (fix_trunc<mode>si2_internal): Likewise.
22369 (fixuns_trunc<mode>si2): Likewise.
22370 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22371 (fctiw<u>z_<mode>_smallint): Likewise.
22372 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
22373 of floating point to 32-bit integer from doing a direct move to
22374 the GPR registers to do a store.
22375 (fctiwz_<mode>): Break long line.
22377 2017-05-05 Bin Cheng <bin.cheng@arm.com>
22379 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
22380 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
22381 (addr_list, addr_offset_valid_p): New.
22382 (split_address_groups): Check offset validity with above function.
22383 (gt-tree-ssa-loop-ivopts.h): Include header file.
22385 2017-05-05 Nathan Sidwell <nathan@acm.org>
22387 * config.gcc (arm*-*-*): Add missing 'fi'.
22389 2017-05-05 Steve Ellcey <sellcey@cavium.com>
22391 * doc/invoke.texi (-fopt-info): Explicitly say order of options
22392 included in -fopt-info does not matter.
22393 * doc/optinfo.texi (-fopt-info): Fix description of default
22394 behavour. Explicitly say order of options included in -fopt-info
22397 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22399 * config.gcc: Allow combinations of aprofile and rmprofile values for
22400 --with-multilib-list.
22401 * config/arm/t-multilib: New file.
22402 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
22403 variables. Remove setting of ISA and floating-point ABI in
22404 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
22405 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
22406 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
22407 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
22409 * config/arm/t-rmprofile: Likewise except for the matches changes.
22410 * doc/install.texi (--with-multilib-list): Document the combination of
22411 aprofile and rmprofile values and warn about pitfalls in doing that.
22413 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22415 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
22416 (movdi_aarch64): Likewise.
22418 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22420 PR tree-optimization/80632
22421 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
22423 (build_arrays): Initialize it for virtual phis.
22424 (fix_phi_nodes): Use it for virtual phis.
22426 PR tree-optimization/80558
22427 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
22428 [x, y] op z into [x op, y op z] for op & or | if conditions
22431 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
22432 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22435 * config/arm/arm.md (use_literal_pool): Remove.
22436 (64-bit immediate split): No longer takes cost into consideration
22437 if arm_disable_literal_pool is enabled.
22438 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
22439 used when arm_disable_literal_pool is enabled.
22440 (arm_max_const_double_inline_cost): Remove use of
22441 arm_disable_literal_pool.
22442 (push_minipool_fix): Add assert.
22443 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
22444 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
22445 (no_literal_pool_sf_immediate): New.
22447 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22449 PR tree-optimization/80613
22450 * tree-ssa-dce.c (propagate_necessity): Remove cases for
22451 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
22453 2017-05-05 Richard Biener <rguenther@suse.de>
22455 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
22457 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
22459 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
22460 of this flag from insn conditions due to removal from r247495.
22462 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22464 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
22466 (arm_early_store_addr_dep_ptr): Likewise.
22467 * config/arm/aarch-common-protos.h
22468 (arm_early_load_addr_dep_ptr): Add prototype.
22469 (arm_early_store_addr_dep_ptr): Likewise.
22470 * config/arm/cortex-a53.md: Add new bypasses.
22472 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22474 * tree.c (next_type_uid): Change type to unsigned.
22475 (type_hash_canon): Decrement back next_type_uid if
22476 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
22477 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
22480 2017-05-04 Martin Sebor <msebor@redhat.com>
22482 * builtins.c: Fix a trivial typo in a comment.
22484 PR middle-end/79234
22485 * builtins.c (check_sizes): Adjust to handle reading past the end.
22486 Avoid printing excessive upper bound of ranges. Use %E to print
22487 tree nodes instead of converting them to %wu.
22488 (expand_builtin_memchr): New function.
22489 (compute_dest_size): Rename...
22490 (compute_objsize): ...to this.
22491 (expand_builtin_memcpy): Adjust.
22492 (expand_builtin_mempcpy): Adjust.
22493 (expand_builtin_strcat): Adjust.
22494 (expand_builtin_strcpy): Adjust.
22495 (check_strncat_sizes): Adjust.
22496 (expand_builtin_strncat): Adjust.
22497 (expand_builtin_strncpy): Adjust and simplify.
22498 (expand_builtin_memset): Adjust.
22499 (expand_builtin_bzero): Adjust.
22500 (expand_builtin_memcmp): Adjust.
22501 (expand_builtin): Handle memcmp.
22502 (maybe_emit_chk_warning): Check strncat just once.
22504 2017-05-04 Martin Sebor <msebor@redhat.com>
22506 PR preprocessor/79214
22507 PR middle-end/79222
22508 PR middle-end/79223
22509 * builtins.c (check_sizes): Add inlining context and issue
22510 warnings even when -Wno-system-headers is set.
22511 (check_strncat_sizes): Same.
22512 (expand_builtin_strncat): Same.
22513 (expand_builtin_memmove): New function.
22514 (expand_builtin_stpncpy): Same.
22515 (expand_builtin): Handle memmove and stpncpy.
22517 2017-05-04 Bin Cheng <bin.cheng@arm.com>
22519 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
22520 which is not used any more.
22522 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22524 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
22526 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22528 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
22529 (cortexa53_tunings): Likewise.
22530 (cortexa57_tunings): Likewise.
22531 (cortexa72_tunings): Likewise.
22532 (cortexa73_tunings): Likewise.
22534 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22536 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
22537 Set loop alignment to 8.
22539 2017-05-04 Martin Sebor <msebor@redhat.com>
22541 PR translation/80280
22542 * builtins.c (expand_builtin_object_size): Add missing quoting to
22543 %D and like directives.
22544 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
22545 (hsa_type_for_tree_type): Same.
22546 (verify_function_arguments): Same.
22547 * symtab.c (symbol_table::change_decl_assembler_name): Same.
22548 * varasm.c (get_section): Same.
22551 2017-05-04 Martin Sebor <msebor@redhat.com>
22553 PR translation/80280
22554 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
22556 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22558 * config/aarch64/aarch64.c (generic_addrcost_table):
22559 Change HI/TI mode setting.
22561 2017-05-04 Martin Jambor <mjambor@suse.cz>
22563 PR tree-optimization/80622
22564 * tree-sra.c (comes_initialized_p): New function.
22565 (build_accesses_from_assign): Only set write lazily when
22566 comes_initialized_p is false.
22567 (analyze_access_subtree): Use comes_initialized_p.
22568 (propagate_subaccesses_across_link): Assert !comes_initialized_p
22569 instead of testing for PARM_DECL.
22571 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22573 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
22574 constraint on operand 0 to allow more general addressing modes.
22575 Adjust output template.
22576 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
22578 * config/aarch64/aarch64-protos.h
22579 (aarch64_address_valid_for_prefetch_p): Declare prototype.
22580 * config/aarch64/constraints.md (Dp): New address constraint.
22581 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
22584 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
22586 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
22587 update use of estimate_ipcp_clone_size_and_time.
22588 (estimate_local_effects): Update use of
22589 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
22590 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
22591 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
22592 Return nonspecialized time.
22594 2017-05-04 Richard Biener <rguenther@suse.de>
22596 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
22597 for the last VUSE which def dominates the PHI. Directly call
22599 (get_continuation_for_phi_1): Remove.
22601 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
22603 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
22604 to explain the use of truncating division. Cap the number of
22605 iterations to the maximum given by nb_iterations_upper_bound,
22608 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22610 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
22611 * configure: Regenerate.
22612 * config.in: Regenerate.
22613 * config/i386/driver-mingw32.c: new file.
22614 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
22615 * config.host: Link driver-mingw32.o on MinGW host.
22616 * doc/install.texi: Document new --enable-mingw-wildcard configure
22619 2017-05-04 Marek Polacek <polacek@redhat.com>
22621 PR tree-optimization/80612
22622 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
22624 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22625 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
22627 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
22628 (movt splitter): Likewise.
22629 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
22630 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
22631 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
22632 block for Thumb-1 with MOVT.
22633 (thumb2_legitimate_address_p): Move code block ...
22634 (can_avoid_literal_pool_for_label_p): ... into this new function.
22635 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
22637 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
22638 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
22639 "M-profile targets with the MOVT instruction".
22641 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22643 * config/arm/arm-builtins.c (arm_init_builtins): Rename
22644 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
22645 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
22647 2017-05-04 Martin Liska <mliska@suse.cz>
22649 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
22650 variable cond_code.
22652 2017-05-04 Richard Biener <rguenther@suse.de>
22654 * tree.c (array_at_struct_end_p): Handle arrays at struct
22655 end with flexarrays more conservatively. Refactor and treat
22656 arrays of arrays or aggregates more strict. Fix
22657 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
22658 * tree.c (array_at_struct_end_p): Adjust prototype.
22659 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
22660 * gimple-fold.c (get_range_strlen): Likewise.
22661 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
22663 2017-05-04 Richard Biener <rguenther@suse.de>
22665 PR tree-optimization/31130
22666 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
22668 (supports_overflow_infinity): Likewise.
22669 (is_negative_overflow_infinity): Likewise.
22670 (is_positive_overflow_infinity): Likewise.
22671 (is_overflow_infinity): Likewise.
22672 (stmt_overflow_infinity): Likewise.
22673 (overflow_infinity_range_p): Likewise.
22674 (usable_range_p): Remove as always returning true.
22675 (make_overflow_infinity): Remove.
22676 (negative_overflow_infinity): Likewise.
22677 (positive_overflow_infinity): Likewise.
22678 (avoid_overflow_infinity): Likewise.
22679 (set_value_range): Adjust accordingly.
22680 (set_value_range_to_nonnegative): Likewise, remove now unused
22681 overflow_infinity arg.
22682 (vrp_operand_equal_p): Adjust.
22683 (update_value_range): Likewise.
22684 (range_int_cst_singleton_p): Likewise.
22685 (operand_less_p): Likewise.
22686 (compare_values_warnv): Likewise.
22687 (extract_range_for_var_from_comparison_expr): Likewise.
22688 (vrp_int_const_binop): Likewise.
22689 (zero_nonzero_bits_from_vr): Likewise.
22690 (extract_range_from_multiplicative_op_1): Likewise.
22691 (extract_range_from_binary_expr_1): Likewise.
22692 (extract_range_from_unary_expr): Likewise.
22693 (extract_range_from_comparison): Likewise.
22694 (extract_range_basic): Likewise.
22695 (adjust_range_with_scev): Likewise.
22696 (compare_ranges): Likewise.
22697 (compare_range_with_value): Likewise.
22698 (dump_value_range): Likewise.
22699 (test_for_singularity): Likewise, remove strict_overflow_p parameter
22701 (simplify_cond_using_ranges): Adjust.
22703 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22705 * brig-builtins.def: Added a builtin for class_f64.
22706 * builtin-types.def: Added a builtin type needed by class_f64.
22708 2017-05-03 Jason Merrill <jason@redhat.com>
22710 * timevar.def: Add TV_CONSTEXPR.
22712 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22714 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
22716 2017-05-03 Martin Jambor <mjambor@suse.cz>
22718 * ipa-prop.c (ipa_update_after_lto_read): Removed.
22719 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
22720 * ipa-cp.c (ipcp_propagate_stage): Do not call
22721 ipa_update_after_lto_read.
22722 * ipa-inline.c (ipa_inline): Likewise.
22724 2017-05-03 Martin Jambor <mjambor@suse.cz>
22726 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
22727 tag. Added a default constructor and a destructor.
22728 (ipa_edge_args_sum_t): New class;
22729 (ipa_edge_args_sum): Declare.
22730 (ipa_edge_args_vector): Remove declaration.
22731 (IPA_EDGE_REF): Use ipa_edge_args_sum.
22732 (ipa_free_edge_args_substructures): Remove declaration.
22733 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
22734 (ipa_edge_args_info_available_for_edge_p): Likewise.
22735 * ipa-prop.c (ipa_edge_args_vector): Removed.
22736 (edge_removal_hook_holder): Likewise.
22737 (edge_duplication_hook_holder): Likewise.
22738 (ipa_edge_args_sum): New variable.
22739 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
22740 ipa_edge_args_vector.
22741 (ipa_free_edge_args_substructures): Likewise.
22742 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
22743 ipa_edge_args_vector.
22744 (ipa_edge_removal_hook): Turned into method
22745 ipa_edge_args_sum_t::remove.
22746 (ipa_edge_duplication_hook): Turned into method
22747 ipa_edge_args_sum_t::duplicate.
22748 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
22749 registering edge hooks.
22750 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
22751 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
22752 ipa_edge_args_sum instead of ipa_edge_args_vector.
22753 * ipa-profile.c (ipa_profile): Likewise.
22755 2017-05-03 Martin Jambor <mjambor@suse.cz>
22757 * symbol-summary.h (function_summary): New method exists.
22758 (function_summary::symtab_removal): Deallocate through release.
22759 (call_summary): New class.
22760 (gt_ggc_mx): New overload.
22761 (gt_pch_nx): Likewise.
22762 (gt_pch_nx): Likewise.
22764 2017-05-03 Jeff Law <law@redhat.com>
22766 PR tree-optimization/78496
22767 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
22768 from simplify_cond_using_ranges. Split off code to walk
22769 backwards through casts into ...
22770 (simplify_cond_using_ranges_2): New function.
22771 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
22772 (execute_vrp): After identifying jump threads, call
22773 simplify_cond_using_ranges_2.
22775 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
22778 * ipa-inline.h (inline_summary): Add ctor.
22779 (create_ggc): Do not use ggc_cleared_alloc.
22781 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
22782 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22784 * gcc.c (handle_braces): Support escaping in switch matching
22786 * doc/invoke.texi (Spec Files): Document it.
22787 Remove superfluous @code markup in items.
22789 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22791 * diagnostic-show-locus.c (struct column_range): New struct.
22792 (get_affected_columns): New function.
22793 (get_printed_columns): New function.
22794 (struct correction): New struct.
22795 (correction::ensure_capacity): New function.
22796 (correction::ensure_terminated): New function.
22797 (struct line_corrections): New struct.
22798 (line_corrections::~line_corrections): New dtor.
22799 (line_corrections::add_hint): New function.
22800 (layout::print_trailing_fixits): Reimplement in terms of the new
22802 (selftest::test_overlapped_fixit_printing): New function.
22803 (selftest::diagnostic_show_locus_c_tests): Call it.
22805 2017-05-03 Nathan Sidwell <nathan@acm.org>
22807 Canonicalize canonical type hashing
22808 * tree.h (type_hash_canon_hash): Declare.
22809 * tree.c (type_hash_list, attribute_hash_list): Move into
22810 type_hash_canon_hash.
22811 (build_type_attribute_qual_variant): Break out hash code calc into
22812 type_hash_canon_hash.
22813 (type_hash_canon_hash): New. Generic type hash computation.
22814 (build_range_type_1, build_array_type_1, build_function_type,
22815 build_method_type_directly, build_offset_type, build_complex_type,
22816 make_vector_type): Call it.
22818 2017-05-03 Richard Biener <rguenther@suse.de>
22820 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
22821 When all DRs have unknown misaligned do not always peel
22822 when there is a store but apply the same costing model as if
22823 there were only loads.
22825 2017-05-03 Richard Biener <rguenther@suse.de>
22828 PR tree-optimization/80492
22829 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22830 compare_base_decls returning dont-know properly.
22832 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
22834 * config/arm/iterators.md (CCSI): New mode iterator.
22835 (arch): New mode attribute.
22836 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
22837 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
22838 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
22839 code iterator for success result mode.
22840 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
22841 the corresponding new insn generators.
22843 2017-05-03 Bin Cheng <bin.cheng@arm.com>
22846 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22847 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22849 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
22851 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
22852 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
22853 (DDR_A): Wrap DDR argument in brackets.
22854 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
22855 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
22856 (DDR_REVERSED_P): Likewise.
22858 2017-05-03 Jakub Jelinek <jakub@redhat.com>
22860 PR tree-optimization/79472
22861 * tree-switch-conversion.c (struct switch_conv_info): Add
22862 contiguous_range and default_case_nonstandard fields.
22863 (collect_switch_conv_info): Compute contiguous_range and
22864 default_case_nonstandard fields, don't clear final_bb if
22865 contiguous_range and only the default case doesn't have the required
22867 (check_all_empty_except_final): Set default_case_nonstandard instead
22868 of failing if contiguous_range and the default case doesn't have empty
22870 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
22871 and only the default case doesn't have the required constants. Skip
22873 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
22874 if default_case_nonstandard.
22875 (build_constructors): Build constant 1 just once. Assert that default
22876 values aren't inserted in between cases if contiguous_range. Skip
22878 (build_arrays): Skip virtual phis.
22879 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
22880 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
22881 Handle virtual phis.
22882 (gen_inbound_check): Handle default_case_nonstandard case.
22883 (process_switch): Adjust check_final_bb caller. Call
22884 gather_default_values with the first non-default case instead of
22885 default case if default_case_nonstandard.
22887 2017-05-02 Nathan Sidwell <nathan@acm.org>
22889 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
22890 check. Fix formatting.
22892 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
22894 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
22895 errors when comparing specialized and unspecialized times.
22897 2017-05-02 David Malcolm <dmalcolm@redhat.com>
22899 * diagnostic-show-locus.c
22900 (layout::should_print_annotation_line_p): Make private.
22901 (layout::print_annotation_line): Make private.
22902 (layout::annotation_line_showed_range_p): Make private.
22903 (layout::show_ruler): Make private.
22904 (layout::print_source_line): Make private. Pass in line and
22905 line_width, rather than calling location_get_source_line. Drop
22907 (layout::print_leading_fixits): New method.
22908 (layout::print_any_fixits): Rename to...
22909 (layout::print_trailing_fixits): ...this, and make private.
22910 Don't print newline fixits.
22911 (diagnostic_show_locus): Move logic for printing one row into...
22912 (layout::print_line): ...this new function. Move the
22913 location_get_source_line call and error-handling from
22914 print_source_line to here. Call print_leading_fixits, and rename
22915 print_any_fixits to print_trailing_fixits.
22916 (selftest::test_fixit_insert_containing_newline): Update now that
22917 newlines are partially supported.
22918 (selftest::test_fixit_insert_containing_newline_2): New test.
22919 (selftest::test_fixit_replace_containing_newline): Update comments.
22920 (selftest::diagnostic_show_locus_c_tests): Call the new test.
22921 * edit-context.c (class added_line): New class.
22922 (class edited_line): Describe newline handling in comment.
22923 (edited_line::actually_edited_p): New method.
22924 (edited_line::print_content): Delete redundant decl.
22925 (edited_line::m_predecessors): New field.
22926 (edited_file::print_content): Call edited_line::print_content.
22927 (edited_file::print_diff): Update to support newlines.
22928 (edited_file::print_diff_hunk): Likewise.
22929 (edited_file::print_run_of_changed_lines): New function.
22930 (edited_file::print_diff_line): Convert to...
22931 (print_diff_line): ...this.
22932 (edited_file::get_effective_line_count): New function.
22933 (edited_line::edited_line): Initialize new field m_predecessors.
22934 (edited_line::~edited_line): Clean up m_predecessors.
22935 (edited_line::apply_fixit): Handle newlines.
22936 (edited_line::get_effective_line_count): New function.
22937 (edited_line::print_content): New function.
22938 (edited_line::print_diff_lines): New function.
22939 (selftest::test_applying_fixits_insert_containing_newline): New
22941 (selftest::test_applying_fixits_replace_containing_newline): New
22943 (selftest::insert_line): New function.
22944 (selftest::test_applying_fixits_multiple_lines): Add example of
22946 (selftest::edit_context_c_tests): Call the new tests.
22948 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22950 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
22951 parameter cand. Update dump information.
22952 (get_computation_cost): Update uses.
22954 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22956 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
22957 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
22958 (get_computation_at, rewrite_use_address): Update use of
22959 get_computation_aff.
22961 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22963 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
22964 (get_computation): Delete.
22965 (get_computation_cost): Implement like get_computation_cost_at.
22966 Use get_computation_at.
22967 (get_computation_cost_at): Delete.
22968 (rewrite_use_nonlinear_expr): Use get_computation_at.
22969 (rewrite_use_compare, remove_unused_ivs): Ditto.
22971 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22973 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
22975 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22977 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
22978 (ivopts_global_cost_for_size): Rename parameter and update uses.
22979 (iv_ca_recount_cost): Update uses.
22980 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
22981 candidates seperately in n_invs and n_cands.
22982 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
22984 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22986 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
22987 (find_inv_vars_cb): New.
22988 (find_depends): Renamed to ...
22989 (find_inv_vars): ... this.
22990 (add_candidate_1, force_var_cost): Call find_inv_vars.
22991 (split_address_cost, determine_group_iv_cost_cond): Ditto.
22993 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22995 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
22996 inv_vars. Add inv_exprs.
22997 (struct iv_cand): Rename depends_on to inv_vars.
22998 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
22999 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
23000 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
23001 (dump_cand): Dump inv_vars.
23002 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
23003 (record_invariant, find_depends, add_candidate_1): Ditto.
23004 (set_group_iv_cost, force_var_cost): Ditto.
23005 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
23006 (get_computation_cost_at, get_computation_cost): Ditto.
23007 (determine_group_iv_cost_generic): Ditto.
23008 (determine_group_iv_cost_address): Ditto.
23009 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
23010 (determine_group_iv_costs): Ditto.
23011 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
23012 (iv_ca_set_remove_invariants): Renamed to ...
23013 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
23014 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
23015 (iv_ca_set_add_invariants): Renamed to ...
23016 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
23017 (iv_ca_set_cp): Use iv_ca_set_add_invs.
23018 (iv_ca_has_deps): Support inv_vars and inv_exprs.
23019 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
23020 (create_new_ivs): Remove useless dump.
23022 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23024 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
23026 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
23027 (iv_ca_set_no_cp, create_new_iv): Ditto.
23029 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23031 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23033 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23035 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
23036 function tree_check2.
23038 2017-05-02 Martin Liska <mliska@suse.cz>
23040 * doc/gcov.texi: Add missing preposition.
23041 * gcov.c (function_info::function_info): Properly fill up
23042 all member variables.
23044 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23046 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
23048 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23050 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
23052 2017-05-02 Martin Liska <mliska@suse.cz>
23055 * lto-streamer-in.c (lto_read_tree_1): Remove
23056 LTO_STREAMER_DEBUG.
23057 * lto-streamer.c (struct tree_hash_entry): Likewise.
23058 (struct tree_entry_hasher): Likewise.
23059 (tree_entry_hasher::hash): Likewise.
23060 (tree_entry_hasher::equal): Likewise.
23061 (lto_streamer_init): Likewise.
23062 (lto_orig_address_map): Likewise.
23063 (lto_orig_address_get): Likewise.
23064 (lto_orig_address_remove): Likewise.
23065 * lto-streamer.h: Likewise.
23066 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
23067 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
23069 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
23071 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
23072 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
23073 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
23074 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
23075 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
23076 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
23077 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
23078 (mm_maskz_sub_ss): New intrinsics.
23079 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23080 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23081 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
23082 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
23083 (__builtin_ia32_subss_mask_round): New builtins.
23084 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23085 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23086 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
23088 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
23089 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23091 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23094 2017-05-02 Martin Jambor <mjambor@suse.cz>
23096 PR tree-optimization/78687
23097 * tree-sra.c (access): New field parent.
23098 (process_subtree_disqualification): New function.
23099 (disqualify_candidate): Call it.
23100 (build_accesses_from_assign): Reset write flag if creating an
23102 (build_access_subtree): Fill in parent field and also prpagate
23103 down grp_write flag.
23104 (create_artificial_child_access): New parameter set_grp_write, set
23105 grp_write to its value.
23106 (propagate_subaccesses_across_link): Also propagate grp_write flag
23108 (propagate_all_subaccesses): Push the closest parent back to work
23109 queue if add_access_to_work_queue returned true.
23111 2017-05-02 Richard Biener <rguenther@suse.de>
23113 * common.opt (fstrict-overflow): Alias negative to fwrapv.
23114 * doc/invoke.texi (fstrict-overflow): Remove all traces of
23115 -fstrict-overflow documentation.
23116 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
23117 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
23118 flag_strict_overflow.
23119 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
23120 * lto-opts.c (lto_write_options): Do not stream it.
23121 * lto-wrapper.c (merge_and_complain): Do not handle it.
23122 * opts.c (default_options_table): Do not set -fstrict-overflow.
23123 (finish_options): Likewise do not clear it when sanitizing.
23124 * simplify-rtx.c (simplify_const_relational_operation): Do not
23125 test flag_strict_overflow.
23127 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
23129 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
23130 using enabled attribute.
23131 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
23132 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
23133 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
23134 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
23135 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
23136 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
23137 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
23138 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
23139 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
23140 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
23142 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
23144 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
23146 2017-05-02 Richard Biener <rguenther@suse.de>
23148 PR tree-optimization/80591
23150 2017-04-10 Richard Biener <rguenther@suse.de>
23152 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23155 2017-05-02 Richard Biener <rguenther@suse.de>
23157 PR tree-optimization/80549
23158 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
23159 (cleanup_tree_cfg_noloop): Create forwarders to known loop
23160 headers if they do not have a preheader.
23162 2017-05-02 Martin Liska <mliska@suse.cz>
23165 * common.opt: Fix typo.
23166 * doc/invoke.texi: Likewise.
23168 2017-05-01 Jan Beulich <jbeulich@suse.com>
23170 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
23171 swapping, add (x,x,m,x,n) alternative.
23173 2017-05-01 Nathan Sidwell <nathan@acm.org>
23175 * calls.c (combine_pending_stack_adjustment_and_call): Remove
23176 unnecessary unadjusted_alignment check.
23178 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23181 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
23183 * gimplify.c (gimplify_cilk_detach): New function.
23184 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
23185 * tree-core.h: Document EXPR_CILK_SPAWN.
23186 * tree.h (EXPR_CILK_SPAWN): Define.
23188 2017-05-01 David Malcolm <dmalcolm@redhat.com>
23190 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
23191 to use new fixit_hint representation, using the "replace" logic.
23192 (get_line_span_for_fixit_hint): Likewise.
23193 (layout::print_any_fixits): Likewise.
23194 (selftest::test_one_liner_many_fixits): Rename to...
23195 (selftest::test_one_liner_many_fixits_1): ...this, and update
23196 comment and expected output to reflect that the multiple fix-it
23197 hints are now consolidated into one insertion.
23198 (selftest::test_one_liner_many_fixits_2): New test.
23199 (selftest::test_diagnostic_show_locus_one_liner): Update for
23201 (selftest::test_fixit_consolidation): Update for fix-it API
23203 * diagnostic.c (print_parseable_fixits): Likewise.
23204 * edit-context.c (edited_line::m_line_events): Convert from
23205 auto_vec <line_event *> to auto_vec <line_event>.
23206 (class line_event): Convert from abstract base class to a concrete
23207 class, taking over the role of replace_event.
23208 (class insert_event): Delete.
23209 (class replace_event): Rename to class line_event. Convert to
23211 (edit_context::add_fixits): Reimplement.
23212 (edit_context::apply_insert): Delete.
23213 (edit_context::apply_replace): Rename to...
23214 (edit_context::apply_fixit): ...this. Convert to half-open range.
23215 (edited_file::apply_insert): Delete.
23216 (edited_file::apply_replace): Rename to...
23217 (edited_file::apply_fixit): ...this.
23218 (edited_line::~edited_line): Drop deletion of events.
23219 (edited_line::apply_insert): Delete.
23220 (edited_line::apply_replace): Rename to...
23221 (edited_line::apply_fixit): ...this. Convert to half-open range.
23222 Update for change to type of m_line_events.
23223 * edit-context.h (edit_context::apply_insert): Delete.
23224 (edit_context::apply_replace): Rename to...
23225 (edit_context::apply_fixit): ...this.
23227 2017-05-01 Martin Sebor <msebor@redhat.com>
23229 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
23232 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
23235 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
23236 __get_cpuid_max returns 0.
23237 (__get_cpuid_count): Ditto.
23239 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
23241 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
23242 replacement expression is another instance of one of its arguments.
23244 2017-05-01 Jakub Jelinek <jakub@redhat.com>
23247 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
23248 check for stack push/pop autoinc.
23249 * config/i386/i386.c (ix86_agi_dependent): Return false
23250 if the only reason why modified_in_p returned true is that
23251 addr is SP based and set_insn is a push or pop.
23253 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
23255 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
23258 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
23261 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
23262 (account_size_time): Use two predicates - exec_pred and
23264 (evaluate_conditions_for_known_args): Compute both clause and
23266 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
23267 (inline_summary_t::duplicate): Update.
23268 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
23270 (compute_inline_parameters): Likewise.
23271 (estimate_edge_size_and_time): Update caluclation of time.
23272 (estimate_node_size_and_time): Compute both time and nonspecialized
23274 (estimate_ipcp_clone_size_and_time): Update.
23275 (inline_merge_summary): Update.
23276 (do_estimate_edge_time): Update.
23277 (do_estimate_edge_size): Update.
23278 (do_estimate_edge_hints): Update.
23279 (inline_read_section, inline_write_summary): Stream both new predicates.
23280 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
23282 (compute_inlined_call_time): Cleanup.
23283 (big_speedup_p): Update.
23284 (edge_badness): Update.
23285 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
23286 (size_time_entry): Replace predicate by exec_predicate and
23287 nonconst_predicate.
23288 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
23289 (estimate_edge_time): Return also nonspec_time.
23290 (reset_edge_growth_cache): Update.
23292 2017-04-29 Jakub Jelinek <jakub@redhat.com>
23294 PR rtl-optimization/80491
23295 * ifcvt.c (noce_process_if_block): When looking for x setter
23296 with missing else_bb, don't check only the insn right before
23297 cond_earliest, but look for the last insn that x is modified in
23298 within the same bb.
23300 PR rtl-optimization/80491
23301 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
23303 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
23305 PR tree-optimization/80487
23306 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
23308 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23310 PR tree-optimization/79697
23311 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
23312 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
23313 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
23315 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
23316 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
23318 2017-04-28 Martin Sebor <msebor@redhat.com>
23320 PR tree-optimization/80523
23321 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
23322 (init_target_to_host_charmap, target_to_host, target_strtol10): New
23324 (maybe_warn, format_directive, parse_directive): Use new functions.
23325 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
23327 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
23329 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
23331 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
23333 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
23334 target_header_dir): Set correctly.
23335 * configure: Regenerated.
23336 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23337 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23338 instead of SYSTEM_HEADER_DIR.
23340 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
23342 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
23343 (estimate_local_effects): Likewise.
23344 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
23345 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
23346 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
23347 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
23348 do_estimate_edge_time, estimate_edge_time): Likewise.
23349 * ipa-inline-analysis.c (estimate_node_size_and_time,
23350 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
23351 (estimate_time_after_inlining): Remove.
23353 2017-04-28 Martin Liska <mliska@suse.cz>
23355 * doc/gcov.texi: Enhance documentation of gcov.
23357 2017-04-28 Martin Liska <mliska@suse.cz>
23359 * doc/gcov.texi: Sort options in alphabetic order.
23360 * doc/gcov-dump.texi: Likewise.
23361 * doc/gcov-tool.texi: Likewise.
23362 * gcov.c (print_usage): Likewise.
23363 * gcov-dump.c (print_usage): Likewise.
23364 * gcov-tool.c (print_merge_usage_message): Likewise.
23365 (print_rewrite_usage_message): Likewise.
23366 (print_overlap_usage_message): Likewise.
23368 2017-04-28 Martin Liska <mliska@suse.cz>
23370 PR gcov-profile/53915
23371 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
23373 2017-04-28 Martin Liska <mliska@suse.cz>
23375 PR gcov-profile/79891
23376 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
23377 is marked by compiler as living on a line.
23378 (get_cycles_count): Remove usage of the union.
23379 (output_intermediate_file): Likewise.
23380 (find_source): Fix GNU coding style.
23381 (accumulate_line_counts): Remove old non-all block mode.
23382 (output_lines): Remove usage of the union.
23383 * profile.c (output_location): Include all BBs, even if
23384 belonging to a same line (and file) as a previous BB.
23386 2017-04-28 Martin Liska <mliska@suse.cz>
23388 * gcov.c (process_args): Handle new argument 'w'.
23389 (read_graph_file): Assign ID to BBs.
23390 (output_branch_count): Display BB # if verbose flag is set.
23391 (output_lines): Likewise for arcs.
23392 (print_usage): Add '--verbose' option help.
23393 * doc/gcov.texi: Document --verbose (-w) option.
23395 2017-04-28 Martin Liska <mliska@suse.cz>
23397 * gcov.c (struct block_location_info): New struct.
23398 (process_file): Fill up the new structure.
23399 (read_graph_file): Replace usage of encoding by the newly added
23401 (add_line_counts): Likewise.
23402 (accumulate_line_counts): Remove usage of the union.
23403 (function_info::function_info): New function.
23404 (function_info::~function_info): Likewise.
23405 (process_file): Call delete instead of release_function.
23406 (release_function): Release the function.
23407 (release_structures): Call delete instead of release_function.
23408 (solve_flow_graph): Replace usage of num_blocks.
23409 (find_exception_blocks): Likewise.
23410 (output_lines): Fix GNU coding style.
23412 2017-04-28 Martin Liska <mliska@suse.cz>
23415 * coverage.c (coverage_remove_note_file): New function.
23416 * coverage.h: Declare the function.
23417 * toplev.c (finalize): Clean if an error has been seen.
23419 2017-04-28 Martin Liska <mliska@suse.cz>
23421 PR gcov-profile/80031
23422 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
23423 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
23424 * gcov.c (read_graph_file): Read just number of blocks.
23425 * profile.c (branch_prob): Do not stream 0 flags per a basic
23428 2017-04-28 Martin Liska <mliska@suse.cz>
23430 * gcov-dump.c (tag_*): Add new argument to declarations.
23431 (dump_gcov_file): Likewise.
23432 (tag_blocks): Add and use new argument depth.
23433 (tag_arcs): Likewise.
23434 (tag_lines): Likewise.
23435 (tag_counters): Likewise.
23436 (tag_summary): Likewise.
23437 (dump_working_sets): Use depth to do a proper indentation.
23439 2017-04-28 Jakub Jelinek <jakub@redhat.com>
23442 * cgraph.h (symtab_node::debug_symtab): No longer inline.
23443 * symtab.c (symtab_node::debug_symtab): Move definition here.
23445 2017-04-28 Richard Biener <rguenther@suse.de>
23447 * lto-streamer.h (LTO_major_version): Bump to 7.
23449 2017-04-28 Richard Biener <rguenther@suse.de>
23451 * tree-vrp.c (assert_info): New struct.
23452 (add_assert_info): New helper.
23453 (register_edge_assert_for_2): Refactor to add asserts to a vector
23455 (register_edge_assert_for_1): Likewise.
23456 (register_edge_assert_for): Likewise.
23457 (finish_register_edge_assert_for): New helper actually registering
23458 asserts where live on edge.
23459 (find_conditional_asserts): Adjust.
23460 (find_switch_asserts): Likewise.
23461 (evrp_dom_walker::try_find_new_range): Generalize.
23462 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
23464 2017-04-27 Marek Polacek <polacek@redhat.com>
23467 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
23468 arg10 and arg11 to itype.
23470 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
23472 * doc/extend.texi (Object Size Checking): Improve grammar.
23474 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
23477 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
23478 that the logic for permitting reciprocal estimates matches that
23481 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23484 * tree.c (type_cache_hasher::equal): Only compare
23485 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
23486 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
23487 non-aggregate element types.
23488 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
23489 about the flag on ARRAY_TYPEs in the comment, formatting fix.
23491 2017-04-27 Richard Biener <rguenther@suse.de>
23493 PR middle-end/80533
23494 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
23495 stripping ARRAY_REFs from MEM_EXPR make sure we're not
23496 keeping a reference to a trailing array.
23498 2017-04-27 Richard Biener <rguenther@suse.de>
23500 PR middle-end/80539
23501 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
23502 being in loop-closed SSA form conservatively.
23503 (chrec_fold_multiply_poly_poly): Likewise.
23505 2017-04-27 Tamar Christina <tamar.christina@arm.com>
23507 PR middle-end/79665
23508 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
23509 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
23511 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23514 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
23515 (aarch64_function_arg_alignment): Return unsigned int again, but still
23516 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
23517 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23518 Don't emit -Wpsabi note.
23519 (aarch64_function_arg_boundary): Likewise.
23520 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23523 2017-04-26 Nathan Sidwell <nathan@acm.org>
23525 * tree.h (crc32_unsigned_n): Declare.
23526 (crc32_unsigned, crc32_unsigned): Make inline.
23527 * tree.c (crc32_unsigned_bits): Replace with ...
23528 (crc32_unsigned_n): ... this.
23529 (crc32_unsigned, crc32_byte): Remove.
23530 (crc32_string): Remove unnecessary braces.
23532 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23534 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
23535 * ipa-inline-analysis.c (MAX_TIME): Remove.
23536 (account_size_time): Use sreal for time.
23537 (dump_inline_summary): Update.
23538 (estimate_function_body_sizes): Update.
23539 (estimate_edge_size_and_time): Update.
23540 (estimate_calls_size_and_time): Update.
23541 (estimate_node_size_and_time): Update.
23542 (inline_merge_summary): Update.
23543 (inline_update_overall_summary): Update.
23544 (estimate_time_after_inlining): Update.
23545 (inline_read_section): Update.
23546 (inline_write_summary): Update.
23547 * ipa-inline.c (compute_uninlined_call_time): Update.
23548 (compute_inlined_call_time): Update.
23549 (recursive_inlining): Update.
23550 (inline_small_functions): Update.
23551 (dump_overall_stats): Update.
23552 * ipa-inline.h: Include sreal.h.
23553 (size_time_entry): Turn time to sreal.
23554 (inline_summary): Turn self_time nad time to sreal.
23556 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23558 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
23560 (sreal::stream_out, sreal::stream_in): New.
23561 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
23563 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23565 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
23568 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
23571 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
23573 (dimode_scalar_chain::compute_convert_gain): Ditto.
23574 (dimode_scalar_chain::make_vector_copies): Ditto.
23575 (dimode_scalar_chain::convert_reg): Ditto.
23576 (dimode_scalar_chain::convert_insn): Ditto.
23577 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
23578 (VI248_AVX512BW_1): New mode iterator.
23579 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
23580 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
23583 2017-04-25 Martin Sebor <msebor@redhat.com>
23585 PR tree-optimization/80497
23586 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
23587 constants are representable in HOST_WIDE_INT.
23588 (parse_directive): Ditto.
23590 2017-04-25 Martin Sebor <msebor@redhat.com>
23593 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
23594 (new_zero_array): Adjust signature.
23595 (dom_info::dom_init): Used unsigned rather that size_t.
23596 (dom_info::dom_info): Same.
23598 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23599 Jakub Jelinek <jakub@redhat.com>
23602 * config/arm/arm.c: Include gimple.h.
23603 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23604 returns negative, increment ncrn only if it returned positive.
23605 (arm_needs_doubleword_align): Return int instead of bool,
23606 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
23607 members, but if there is any such non-FIELD_DECL
23608 > PARM_BOUNDARY aligned decl, return -1 instead of false.
23609 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23610 returns negative, increment nregs only if it returned positive.
23611 (arm_setup_incoming_varargs): Likewise.
23612 (arm_function_arg_boundary): Emit -Wpsabi note if
23613 arm_needs_doubleword_align returns negative, return
23614 DOUBLEWORD_ALIGNMENT only if it returned positive.
23616 2017-04-25 Marek Polacek <polacek@redhat.com>
23619 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23620 first argument to type.
23622 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
23625 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
23626 type checks to test for compatibility instead of equality.
23628 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23629 Jakub Jelinek <jakub@redhat.com>
23632 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
23634 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
23635 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
23636 the alignment computation, but return their maximum in warn_alignment.
23637 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23638 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
23640 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
23641 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23644 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23646 * config/arc/simdext.md (dmpyh): Fix typo.
23648 2017-04-25 Richard Biener <rguenther@suse.de>
23650 PR tree-optimization/80492
23651 * alias.c (compare_base_decls): Handle registers with asm
23652 specification conservatively.
23653 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
23654 compare_base_decls returning dont-know properly.
23656 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23658 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
23659 (legitimate_offset_address_p): New function.
23660 (arc_legitimate_address_p): Use above function.
23662 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23664 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
23666 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23668 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
23669 ACCH registers whenever they are available.
23671 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23673 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
23674 double regs fix when not used.
23676 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23678 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
23680 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
23681 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
23683 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23685 * config/arc/arc.c (arc_output_addsi): Check for h-register class
23686 when emitting short ADD instructions.
23688 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23690 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
23692 (cmpsi_cc_c_insn): Likewise.
23693 (cbranchsi4_scratch): Compute proper instruction length using
23694 compact_hreg_operand.
23695 * config/arc/predicates.md (compact_hreg_operand): New predicate.
23697 2017-04-25 Richard Biener <rguenther@suse.de>
23699 PR middle-end/80509
23700 * passes.c (pass_manager::pass_manager): Initialize
23701 m_name_to_pass_map.
23703 2017-04-25 Richard Biener <rguenther@suse.de>
23705 PR tree-optimization/79201
23706 * tree-ssa-sink.c (statement_sink_location): Handle calls.
23708 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23711 * config/s390/vector.md: Split MEM->GPR vector moves for
23712 non-s_operand addresses.
23714 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23717 * config/s390/predicates.md (reload_const_wide_int_operand): New
23719 * config/s390/s390.md ("movti"): Remove d/P alternative.
23720 ("movti_bigconst"): New pattern definition.
23722 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23725 * s390-protos.h (s390_expand_cs_hqi): Removed.
23726 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
23727 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
23728 modes as well as CCZ1mode and CCZmode.
23729 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
23730 signature of s390_emit_compare_and_swap.
23731 (s390_expand_cs_hqi): Likewise, make static.
23732 (s390_expand_cs_tdsi): Generate an explicit compare before trying
23733 compare-and-swap, in some cases.
23734 (s390_expand_cs): Wrapper function.
23735 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
23737 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
23738 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
23739 patterns for small and large integers. Forbid symref memory operands.
23740 Move expander to s390.c. Require cc register.
23741 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
23742 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
23743 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
23744 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
23745 symref memory operands. Remove CC mode and call s390_match_ccmode
23747 ("atomic_exchange<mode>"): Allow and implement all integer modes.
23749 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23751 * config/s390/s390.md (define_peephole2): New peephole to help
23752 combining the load-and-test pattern with volatile memory.
23754 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23756 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
23757 with CCZmode for TARGET_Z196.
23759 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23761 PR rtl-optimization/80501
23762 * combine.c (make_compound_operation_int): Set subreg_code to SET
23763 even for AND with mask of the sign bit of mode.
23765 PR rtl-optimization/80500
23766 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
23767 sum's initial value.
23769 2017-04-25 Julian Brown <julian@codesourcery.com>
23770 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23772 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
23774 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
23776 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
23778 2017-04-25 Julian Brown <julian@codesourcery.com>
23779 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23781 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
23782 (thunderx2t99_sha): New Reservation.
23784 2017-04-25 Julian Brown <julian@codesourcery.com>
23785 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23787 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
23788 type for 1-element load.
23790 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
23792 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
23794 2017-04-24 Martin Jambor <mjambor@suse.cz>
23796 PR tree-optimization/80293
23797 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
23798 char arrays not totally scalarizable if it is false.
23799 (analyze_all_variable_accesses): Pass correct value in the new
23800 parameter. Add a statistics counter.
23802 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
23804 PR middle-end/79931
23805 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
23807 2017-04-24 Richard Biener <rguenther@suse.de>
23809 PR tree-optimization/80494
23810 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
23811 out for complex types.
23813 2017-04-24 Richard Biener <rguenther@suse.de>
23815 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
23816 * tree-ssa-sccvn.c (print_scc): Print SCC size.
23817 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
23818 (DFS): Adjust and never fail.
23819 (sccvn_dom_walker::fail): Remove.
23820 (sccvn_dom_walker::before_dom_children): Adjust.
23821 (run_scc_vn): Likewise and never fail.
23822 * tree-ssa-pre.c (pass_pre::execute): Adjust.
23823 (pass_fre::execute): Likewise.
23825 2017-04-24 Richard Biener <rguenther@suse.de>
23827 PR tree-optimization/79725
23828 * tree-ssa-sink.c (statement_sink_location): Return whether
23829 failure reason was zero uses. Move that check later.
23830 (sink_code_in_bb): Deal with zero uses by removing the stmt
23833 2017-04-24 Richard Biener <rguenther@suse.de>
23836 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
23837 pointer-based references.
23839 2017-04-24 Richard Biener <rguenther@suse.de>
23842 * pass_manager.h (pass_manager::operator new): Remove.
23843 (pass_manager::operator delete): Likewise.
23844 * passes.c (pass_manager::operator new): Remove.
23845 (pass_manager::operator delete): Likewise.
23846 (pass_manager::pass_manager): Zero individual pass members.
23848 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
23851 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
23852 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
23853 Check "XEXP (src, 1)" operand here.
23854 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
23855 Check "XEXP (src, 1)" operand here.
23856 (dimode_scalar_chain::make_vector_copies): Detect count register
23857 of a shift instruction. Zero extend count register from QImode
23858 to DImode to satisfy vector shift pattern count operand predicate.
23859 Substitute vector shift count operand with a DImode copy.
23860 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
23863 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
23865 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
23866 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23867 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
23868 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23869 (UNSPEC_NOREX_MEM): Remove definition.
23871 2017-04-21 Richard Biener <rguenther@suse.de>
23873 PR tree-optimization/79547
23874 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23875 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
23876 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
23877 without any constraints.
23879 2017-04-21 Richard Biener <rguenther@suse.de>
23881 PR tree-optimization/78847
23882 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
23884 2017-04-21 Richard Biener <rguenther@suse.de>
23886 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
23887 (build_distinct_type_copy): Likewise.
23888 (build_variant_type_copy): Likewise.
23889 * tree.c (build_qualified_type): Pass down mem-stat info.
23890 (build_distinct_type_copy): Likewise.
23891 (build_variant_type_copy): Likewise.
23893 2017-04-21 Richard Biener <rguenther@suse.de>
23895 PR tree-optimization/80237
23896 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
23898 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
23899 for a simplified result.
23901 2016-04-21 Richard Biener <rguenther@suse.de>
23903 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
23904 sth as strict as a simple_iv but a chrec without symbols and an
23905 operand defined in the loop we are peeling (and not some subloop).
23906 (propagate_constants_for_unrolling): Propagate all constants.
23908 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23911 * config/i386/i386.c (print_reg): Remove assert for disalowed
23912 regno values, call output_operand_lossage instead.
23914 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23917 * config/i386/constraints.md (Yc): New register constraint.
23918 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
23919 Use Yc constraint for alternative 2 of operand 0. Remove
23920 preferred_for_speed attribute.
23922 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
23924 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
23925 lastprivate clauses in SIMT case.
23927 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
23929 * doc/invoke.texi (-Wextra-semi): Document new warning option.
23931 2017-04-20 Richard Biener <rguenther@suse.de>
23933 PR tree-optimization/57796
23934 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
23935 as N scalar stores.
23936 (vect_model_load_cost): Cost gathers as N scalar loads.
23938 2017-04-20 Richard Biener <rguenther@suse.de>
23940 * ggc-page.c (ggc_allocated_p): Rename to ...
23941 (safe_lookup_page_table_entry): ... this and return the lookup
23943 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
23945 2017-04-20 Richard Biener <rguenther@suse.de>
23947 PR tree-optimization/80453
23948 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
23949 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
23950 from the conditions.
23951 (vn_phi_eq): Pass them down.
23952 (vn_phi_lookup): Record them.
23953 (vn_phi_insert): Likewise.
23955 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
23957 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
23958 uninitialized variable warning to avoid buffer overrun.
23960 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
23963 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
23964 is suppressed for '{ 0 }' in C.
23966 2017-04-20 Jakub Jelinek <jakub@redhat.com>
23968 * BASE-VER: Set to 8.0.0.
23970 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
23972 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
23973 priority .init_array and .fini_array section with SECTION_NOTYPE
23976 2017-04-20 Jakub Jelinek <jakub@redhat.com>
23978 PR middle-end/80423
23979 * tree.h (build_array_type): Add typeless_storage default argument.
23980 * tree.c (type_cache_hasher::equal): Also compare
23981 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
23982 (build_array_type): Add typeless_storage argument, set
23983 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
23985 (build_nonshared_array_type): Adjust build_array_type_1 caller.
23986 (build_array_type): Likewise. Add typeless_storage argument.
23988 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
23989 Jakub Jelinek <jakub@redhat.com>
23991 PR tree-optimization/80426
23992 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
23993 operation on symbolic operands, also compute the overflow for the
23994 invariant part when the operation degenerates into a negation.
23996 2017-04-19 Jakub Jelinek <jakub@redhat.com>
23999 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
24000 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
24003 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
24005 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
24008 * config/avr/avr.c (tree.h): Include it.
24009 (cgraph.h): Include it.
24010 (avr_encode_section_info): Don't warn for uninitialized progmem
24011 variable if it's just an alias.
24013 2017-04-19 Richard Biener <rguenther@suse.de>
24016 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
24017 when needed by AutoPGO.
24019 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
24022 * doc/lto.texi: Remove an extra 'that'.
24024 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
24026 PR rtl-optimization/80429
24027 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
24028 are only used in debug insns.
24030 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24031 Vladimir Makarov <vmakarov@redhat.com>
24033 * config/sparc/predicates.md (input_operand): Add comment. Return
24034 true for any memory operand when LRA is in progress.
24035 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
24037 2017-04-18 Jeff Law <law@redhat.com>
24040 * mips.md ({return,simple_return}_internal): Do not overwrite
24043 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24045 PR tree-optimization/80443
24046 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
24047 instead of adding 1, subtract -1 and similarly instead of subtracting
24050 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
24052 PR rtl-optimization/80357
24053 * haifa-sched.c (tmp_bitmap): New variable.
24054 (model_recompute): Handle duplicate use records.
24055 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
24056 (free_global_sched_pressure_data): Free it.
24058 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
24061 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24062 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24063 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24064 instead of SYSTEM_HEADER_DIR.
24066 2017-04-18 Jeff Law <law@redhat.com>
24068 PR middle-end/80422
24069 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
24070 predecessors after walking up the insn chain.
24072 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24075 * dwarf2out.c (modified_type_die): Try harder not to emit internal
24076 sizetype type into debug info.
24078 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24081 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
24082 unneeded test for TARGET_UPPER_REGS_SF.
24083 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
24085 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24088 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
24089 instead of gsi_after_labels.
24091 2017-04-18 Jeff Law <law@redhat.com>
24093 * regcprop.c (maybe_mode_change): Avoid creating copies of the
24097 2017-04-13 Jeff Law <law@redhat.com>
24098 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
24099 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
24101 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
24104 * config/avr/avr.c (intl.h): Include it.
24105 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
24107 2017-04-18 Martin Liska <mliska@suse.cz>
24109 PR gcov-profile/78783
24110 * gcov-tool.c (gcov_output_files): Validate that destination
24111 file is either removed by the tool or by a user.
24113 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24114 Guy Benyei <guybe@mellanox.com>
24116 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
24117 block, and do not negate it, the stored id is already negative.
24119 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24121 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
24123 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
24126 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
24127 masks of options that should be turned off if the VSX vector
24128 options are turned off.
24129 (OTHER_P8_VECTOR_MASKS): Likewise.
24130 (OTHER_VSX_VECTOR_MASKS): Likewise.
24131 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
24132 rs6000_disable_incompatible_switches to validate no type switches
24134 (rs6000_incompatible_switch): New function to disallow turning on
24135 other vector options if -mno-vsx, -mno-power8-vector, or
24136 -mno-power9-vector are specified.
24138 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24140 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
24142 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24144 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
24145 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
24146 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
24147 (ARG_POINTER_CFA_OFFSET): Likewise.
24149 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24151 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
24152 conditions to take advantage of various optimizations.
24154 2017-04-13 Jeff Law <law@redhat.com>
24156 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
24157 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
24158 (zero_extendsidi2_dext): Likewise.
24160 2017-04-13 Jakub Jelinek <jakub@redhat.com>
24163 * fold-const.c (fold_ternary_loc): Revert
24164 use op0 instead of fold_convert_loc (loc, type, arg0) part of
24167 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
24169 PR rtl-optimization/80343
24170 * lra-remat.c (update_scratch_ops): Assign original hard reg to
24171 new scratch pseudo.
24173 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
24176 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
24177 to ubsan_encode_value.
24179 2017-04-13 Jeff Law <law@redhat.com>
24181 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
24182 appearing in DEBUG_INSNs.
24184 2017-04-13 Martin Liska <mliska@suse.cz>
24186 PR gcov-profile/80413
24187 * gcov-io.c (gcov_write_string): Copy to buffer just when
24188 allocated size is greater than zero.
24190 2017-04-13 Jakub Jelinek <jakub@redhat.com>
24193 * dwarf2out.c (decls_for_scope): Ignore declarations of
24194 current_function_decl in BLOCK_NONLOCALIZED_VARS.
24196 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
24199 * ipa-visibility.c (non_local_p): Fix typos.
24200 (localize_node): When localizing symbol in same comdat group,
24201 dissolve the group only when we know external symbols are going
24203 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
24205 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24207 PR tree-optimization/79390
24208 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
24209 order does not result in usable sequence, retry with reversed operand
24215 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
24216 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
24217 op0 instead of fold_convert_loc (loc, type, arg0).
24219 2017-04-12 Jeff Law <law@redhat.com>
24221 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
24222 has a delay slot in the generated code.
24224 * config/cris/cris.md (cris_preferred_reload_class): Return
24225 GENNONACR_REGS rather than GENERAL_REGS.
24227 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24230 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
24231 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
24232 signedness of the result type.
24234 2017-04-12 Richard Biener <rguenther@suse.de>
24235 Jeff Law <law@redhat.com>
24237 PR tree-optimization/80359
24238 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
24239 trim stores to TARGET_MEM_REFs.
24241 2017-04-12 Richard Biener <rguenther@suse.de>
24243 PR tree-optimization/79390
24244 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
24245 threading case even more.
24247 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
24250 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
24251 for quad_address_p for TImode, instead of just not indexed_address.
24253 2017-04-12 Richard Biener <rguenther@suse.de>
24254 Bernd Edlinger <bernd.edlinger@hotmail.de>
24256 PR middle-end/79671
24257 * alias.c (component_uses_parent_alias_set_from): Handle
24258 TYPE_TYPELESS_STORAGE.
24259 (get_alias_set): Likewise.
24260 * tree-core.h (tree_type_common): Add typeless_storage flag.
24261 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
24262 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
24263 for types containing members with TYPE_TYPELESS_STORAGE.
24264 (place_field): Likewise.
24265 (layout_type): Likewise for ARRAY_TYPE.
24266 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
24267 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
24268 TYPE_TYPELESS_STORAGE.
24269 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
24271 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24274 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
24275 first argument to type.
24277 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24281 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
24282 CONST0_RTX (mode) rather than const0_rtx where appropriate.
24283 (rs6000_expand_binop_builtin): Likewise.
24284 (rs6000_expand_ternop_builtin): Likewise; also add missing
24285 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
24286 vshasigma built-ins.
24287 * doc/extend.texi: Document that vec_xxpermdi's third argument
24288 must be a constant.
24290 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
24292 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
24293 Use shift_const cost parameter when calculating gain of STV shifts.
24295 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
24297 PR rtl-optimization/70478
24298 * lra-constraints.c (process_alt_operands): Check memory for
24299 disfavoring memory insn operand.
24301 2017-04-11 Jakub Jelinek <jakub@redhat.com>
24303 PR middle-end/80100
24304 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
24305 left shift in unsigned HOST_WIDE_INT type.
24307 PR rtl-optimization/80385
24308 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
24309 (not (neg X)) into (plus X -1) for complex or non-integral modes.
24312 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
24313 if they have any depend clauses.
24315 2017-04-11 Martin Liska <mliska@suse.cz>
24318 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
24319 * ipa-split.c (split_function): Create a local comdat symbol
24320 if caller is in a comdat group.
24322 2017-04-11 Martin Liska <mliska@suse.cz>
24325 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
24328 2017-04-11 Martin Sebor <msebor@redhat.com>
24330 PR middle-end/80364
24331 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
24332 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
24334 (directive::set_width, directive::set_precision, format_character):
24336 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
24339 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
24342 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
24343 conflict, set target->arch_name instead of target->cpu_name.
24345 2017-04-11 Richard Biener <rguenther@suse.de>
24347 PR tree-optimization/80374
24348 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
24349 build_zero_cst, remove fold_convertible_p check again.
24351 2017-04-11 Martin Liska <mliska@suse.cz>
24354 * ubsan.c (instrument_object_size): Do not instrument register
24357 2017-04-11 Jakub Jelinek <jakub@redhat.com>
24360 * config/i386/i386-builtin-types.def
24361 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
24362 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
24363 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
24364 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
24365 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
24366 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
24367 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
24368 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
24369 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
24370 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
24371 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
24372 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
24373 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
24374 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
24375 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
24376 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
24377 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
24378 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
24379 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
24380 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
24381 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
24382 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
24383 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
24384 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
24385 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
24386 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
24387 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
24388 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
24389 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
24390 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
24391 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
24392 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
24393 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
24394 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
24395 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
24396 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
24397 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
24398 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
24399 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
24400 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
24401 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
24402 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
24403 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
24404 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
24405 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
24406 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
24408 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
24409 flag to second_arg_count, handle 4 argument function type _COUNT
24410 aliases, handle second_arg_count on second argument rather than last.
24412 2017-04-10 Jeff Law <law@redhat.com>
24414 PR tree-optimization/80374
24415 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
24416 record anything if we can not convert integer_zero_node to the
24419 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24422 * config/rs6000/rs6000.c (rs6000_option_override_internal):
24423 Enhance special handling given to the TARGET_P9_MINMAX option in
24424 relation to certain other options.
24426 2017-04-10 Bin Cheng <bin.cheng@arm.com>
24428 PR tree-optimization/80153
24429 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
24430 remove POINTER_PLUS_EXPR's base part directly, rather than through
24433 2017-04-10 Richard Biener <rguenther@suse.de>
24434 Bin Cheng <bin.cheng@arm.com>
24436 PR tree-optimization/80153
24437 * tree-affine.c (aff_combination_to_tree): Get base pointer from
24438 the first element of pointer type aff_tree. Build result expr in
24440 (add_elt_to_tree): Convert to type unconditionally. Remove other
24441 fold_convert calls.
24442 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
24443 (rewrite_use_nonlinear_expr): Check invariant using iv information.
24445 2017-04-10 Richard Biener <rguenther@suse.de>
24447 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24450 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
24452 PR rtl-optimization/70478
24453 * lra-constraints.c (curr_small_class_check): New.
24454 (update_and_check_small_class_inputs): New.
24455 (process_alt_operands): Update curr_small_class_check. Disfavor
24456 alternative insn memory operands. Check available regs for small
24459 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24462 * config/mips/mips.opt (-mvirt): Update description.
24463 * doc/invoke.texi (-mvirt): Likewise.
24465 2017-04-10 Richard Biener <rguenther@suse.de>
24467 PR middle-end/80362
24468 * fold-const.c (fold_binary_loc): Look at unstripped ops when
24469 looking for NEGATE_EXPR in -A / -B to A / B folding.
24471 2017-04-10 Martin Liska <mliska@suse.cz>
24473 PR gcov-profile/80224
24474 * gcov.c (print_usage): Fix usage string.
24475 (get_gcov_intermediate_filename): Remove.
24476 (output_gcov_file): Use both for normal and intermediate format.
24477 (generate_results): Do not initialize special file for
24478 intermediate format.
24480 2017-04-10 Richard Biener <rguenther@suse.de>
24482 PR tree-optimization/80304
24483 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
24486 2017-04-10 Nathan Sidwell <nathan@acm.org>
24489 * config/rs6000/rs6000.c (rs6000_vector_type): New.
24490 (rs6000_init_builtins): Use it.
24492 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24494 * config/arm/arm.md (<mrc>): Add mode to SET source.
24495 (<mrrc>): Likewise.
24497 2017-04-10 Richard Biener <rguenther@suse.de>
24499 PR middle-end/80344
24500 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
24502 2017-04-10 Jakub Jelinek <jakub@redhat.com>
24505 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
24506 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
24507 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
24508 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
24509 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
24510 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
24511 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
24512 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
24513 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
24514 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
24515 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
24516 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
24517 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
24518 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
24519 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
24520 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
24521 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
24522 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
24523 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
24524 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
24525 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
24526 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
24527 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
24529 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
24531 PR rtl-optimization/70478
24532 * lra-constraints.c: Reverse the last patch.
24534 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
24536 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
24537 Add comment for WCHAR_T.
24539 2017-04-08 Martin Liska <mliska@suse.cz>
24542 2017-04-07 Martin Liska <mliska@suse.cz>
24545 * ipa-split.c (split_function): Add function part to a same comdat
24548 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24551 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
24553 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
24555 * rs6000/rs6000.c (vec_load_pendulum): Rename...
24556 (vec_pairing): ...to this.
24557 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
24558 (rs6000_sched_init): Adjust for name change.
24559 (struct rs6000_sched_context): Likewise.
24560 (rs6000_init_sched_context): Likewise.
24561 (rs6000_set_sched_context): Likewise.
24563 2017-04-07 Jakub Jelinek <jakub@redhat.com>
24569 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
24571 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
24572 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
24573 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
24575 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
24577 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
24579 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24581 PR rtl-optimization/70703
24582 * ira-color.c (update_conflict_hard_regno_costs): Use
24583 int64_t instead of HOST_WIDE_INT.
24585 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24587 PR rtl-optimization/70478
24588 * lra-constraints.c (process_alt_operands): Disfavor alternative
24589 insn memory operands.
24591 2017-04-07 Jeff Law <law@redhat.com>
24593 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
24594 CALL and NOTE_INSN_CALL_ARG_LOCATION.
24596 2017-04-07 Martin Liska <mliska@suse.cz>
24599 * config/aarch64/aarch64.c (aarch64_process_target_attr):
24600 Show error message instead of an ICE.
24602 2017-04-07 Martin Liska <mliska@suse.cz>
24605 * ipa-split.c (split_function): Add function part to a same comdat
24608 2017-04-07 Richard Biener <rguenther@suse.de>
24610 PR middle-end/80341
24611 * tree.c (get_unwidened): Also handle ! for_type case for
24613 * convert.c (do_narrow): Split out from ...
24614 (convert_to_integer_1): ... here. Do not pass final truncation
24615 type to get_unwidened for TRUNC_DIV_EXPR.
24617 2017-04-07 Richard Biener <rguenther@suse.de>
24619 * tree-affine.c (wide_int_ext_for_comb): Take type rather
24621 (aff_combination_const): Adjust.
24622 (aff_combination_scale): Likewise.
24623 (aff_combination_add_elt): Likewise.
24624 (aff_combination_add_cst): Likewise.
24625 (aff_combination_convert): Likewise.
24626 (add_elt_to_tree): Likewise. Remove unused argument.
24627 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
24629 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
24631 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
24633 * config/arm/arm.c (arm_default_short_enums): Use
24634 ARM_DEFAULT_SHORT_ENUMS.
24635 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
24637 2017-04-06 Jakub Jelinek <jakub@redhat.com>
24640 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
24641 members with redundant out-of-class redeclaration.
24643 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24646 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
24647 * config/i386/i386.md (*zero_extendsidi2):
24648 Add (?*x,*x) and (?*v,*v) alternatives.
24650 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24653 * config/i386/i386.c (ix86_expand_builtin)
24654 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
24655 mode from insn data. Convert operands to insn operand mode.
24656 Copy operands that don't satisfy insn predicate to a register.
24658 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
24660 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
24663 2017-04-06 Richard Biener <rguenther@suse.de>
24665 PR tree-optimization/80334
24666 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
24667 preserve alignment of accesses.
24669 2017-04-06 Richard Biener <rguenther@suse.de>
24671 PR tree-optimization/80262
24672 * tree-sra.c (build_ref_for_offset): Preserve address-space
24674 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
24675 Drop useless address-space information on MEM_REF offsets.
24677 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
24679 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
24681 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24683 PR rtl-optimization/70703
24684 * ira-color.c (update_conflict_hard_regno_costs): Use
24685 HOST_WIDE_INT instead of long.
24687 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
24690 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
24691 not defined for x86_64 target. Add -mmmx target option when __SSE2__
24693 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
24694 for x86_64 target. Handle -m3dnowa option.
24696 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24698 PR rtl-optimization/70703
24699 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
24700 (update_conflict_hard_regno_costs): Use long instead of unsigned
24701 arithmetic for cost calculation.
24703 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24704 Bernd Edlinger <bernd.edlinger@hotmail.de>
24707 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
24710 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
24713 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
24714 ptr_mode with Pmode throughout.
24715 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
24716 into probe_stack_range and use DImode.
24718 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
24721 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
24722 call_eh_return is true.
24724 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24726 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
24727 Initialize last_match_fntype_index.
24729 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24732 * tree-nvr.c: Include internal-fn.h.
24733 (pass_return_slot::execute): Ignore internal calls without
24736 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24737 Richard Biener <rguenther@suse.de>
24740 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
24741 captures used multiple times, except for the last use.
24742 * generic-match-head.c: Include gimplify.h.
24744 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24746 PR tree-optimization/79390
24747 * target.h (struct noce_if_info): Declare.
24748 * targhooks.h (default_noce_conversion_profitable_p): Declare.
24749 * target.def (noce_conversion_profitable_p): New target hook.
24750 * ifcvt.h (struct noce_if_info): New type, moved from ...
24751 * ifcvt.c (struct noce_if_info): ... here.
24752 (noce_conversion_profitable_p): Renamed to ...
24753 (default_noce_conversion_profitable_p): ... this. No longer
24755 (noce_try_store_flag_constants, noce_try_addcc,
24756 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
24757 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
24758 instead of noce_conversion_profitable_p.
24759 * config/i386/i386.c: Include ifcvt.h.
24760 (ix86_option_override_internal): Don't override
24761 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
24762 (ix86_noce_conversion_profitable_p): New function.
24763 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
24764 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
24765 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
24766 * doc/tm.texi: Regenerated.
24768 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24770 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
24773 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24776 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
24777 instructions for small multiply cores.
24779 2017-04-04 Jeff Law <law@redhat.com>
24781 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
24783 (mips_expand_vec_perm_const): Initialize elements in orig_perm
24784 that are not set by the loop over the elements.
24786 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24789 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
24790 int mode, convert_modes it to mode as unsigned, otherwise use
24791 lowpart_subreg to mode rather than SImode.
24792 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
24793 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
24794 Use DImode instead of SImode for the shift count operand.
24795 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
24798 2017-04-04 Richard Biener <rguenther@suse.de>
24800 PR middle-end/80281
24801 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
24802 arithmetic done for the negate or the plus. Simplify.
24803 (A - (-B) -> A + B): Likewise.
24804 * fold-const.c (split_tree): Make sure to not negate pointers.
24806 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
24808 PR rtl-optimization/60818
24809 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
24810 a compare of comparisons with the thing compared if this results
24811 in a different machine mode.
24813 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
24815 * alias.c (base_alias_check): Fix typo in comment.
24816 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
24817 * cgraphunit.c (symbol_table::compile): Likewise.
24818 * collect2.c (maybe_run_lto_and_relink): Likewise.
24819 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
24820 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
24821 * config/avr/avr.c (avr_map_op_t): Likewise.
24822 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
24823 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
24824 * config/epiphany/epiphany.md (movcc): Likewise.
24825 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
24826 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
24828 * config/mips/mips.c (mips_save_restore_reg): Likewise.
24829 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
24830 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
24831 * config/sh/sh.c (sh_rtx_costs): Likewise.
24832 * fold-const.c (fold_truth_andor): Likewise.
24833 * genautomata.c (collapse_flag): Likewise.
24834 * gengtype.h (struct type::u::s): Likewise.
24835 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
24836 * input.c (FORMAT_AMOUNT): Likewise.
24837 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
24838 (known_aggs_to_agg_replacement_list): Likewise.
24839 * ipa-inline-analysis.c: Likewise.
24840 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
24841 * ipa-polymorphic-call.c
24842 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
24843 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
24844 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
24846 * modulo-sched.c (apply_reg_moves): Likewise.
24847 * omp-expand.c (build_omp_regions_1): Likewise.
24848 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
24849 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
24850 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
24851 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24852 * value-prof.c: Likewise.
24853 * var-tracking.c (val_reset): Likewise.
24855 2017-04-03 Richard Biener <rguenther@suse.de>
24857 PR tree-optimization/80275
24858 * fold-const.c (split_address_to_core_and_offset): Handle
24861 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
24863 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
24864 descriptors is at least equal to that of functions.
24866 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24868 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
24870 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24873 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
24874 (mov<IMOD4:mode>): New expander.
24875 (*mov<IMOD4:mode>_internal): New insn and split pattern.
24877 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
24879 PR rtl-optimization/79405
24880 * fwprop.c (propagations_left): New variable.
24881 (forward_propagate_into): Decrement it.
24882 (fwprop_init): Initialize it.
24883 (fw_prop): If the variable has reached zero, stop propagating.
24884 (fwprop_addr): Ditto.
24886 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24889 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
24890 a FUNCTION_DECL, pass it as decl instead of origin to
24893 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
24895 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
24898 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
24901 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
24902 TARGET_VSX_SMALL_INTEGER.
24904 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24906 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24907 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
24909 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24911 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
24912 extraction from odd-numbered MSA register.
24914 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24916 PR middle-end/80173
24917 * expmed.c (store_bit_field_1): Don't attempt to create
24918 a word subreg out of hard registers wider than word if they
24919 have HARD_REGNO_NREGS of 1 for their mode.
24921 PR middle-end/80163
24922 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
24923 conversions to integer types wider than word and pointer.
24926 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
24927 (rtx_equal_for_cselib_p): Pass 0 to it.
24928 * cselib.c (cselib_hasher::equal): Likewise.
24929 (rtx_equal_for_cselib_1): Add depth argument. If depth
24930 is 128, don't look up VALUE locs and punt. Increment
24931 depth in recursive calls when walking VALUE locs.
24933 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
24935 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
24936 (make_gcov_file_name): Use the canonical path name for generating
24938 (read_line): Fix handling of files with ascii null bytes.
24940 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
24942 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
24943 to initialise a vector register instead
24944 of using a const_int.
24946 2017-03-30 Jakub Jelinek <jakub@redhat.com>
24948 PR translation/80189
24949 * gimplify.c (omp_default_clause): Use %qs instead of %s in
24950 diagnostic messages.
24952 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
24955 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
24956 (dfp_diex_<mode>): Update mode of operand 1.
24957 * doc/extend.texi (dxex, dxexq): Document change to return type.
24958 (diex, diexq): Document change to argument type.
24960 2017-03-30 Martin Jambor <mjambor@suse.cz>
24963 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
24964 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
24965 it reflects the signature changes performed at the callee side.
24966 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
24967 to cgraph_build_function_type_skip_args.
24968 (build_function_decl_skip_args): Adjust call to the above function.
24970 2017-03-30 Jakub Jelinek <jakub@redhat.com>
24973 * config/i386/sse.md
24974 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
24975 register as dest whenever it is a MEM not rtx_equal_p to the
24976 corresponding dup operand, and when forcing into reg move the
24977 reg into the memory afterwards.
24978 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
24979 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
24980 for the force_reg mode.
24981 (avx512vl_vextractf128<mode>): Use register as dest either
24982 always when a MEM, or when it is a MEM not rtx_equal_p to the
24983 corresponding dup operand, or even not when it is a CONST_VECTOR
24984 depending on the mode and lo vs. hi.
24985 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
24987 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
24988 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
24989 Likewise. Require that operands[2] is even.
24990 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
24991 Remove extraneous parens. Require that operands[2] is a multiple
24993 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
24994 operands[0] is a MEM if <mask_applied>, the predicates/constraints
24995 disallow memory then.
24997 2017-03-30 Richard Biener <rguenther@suse.de>
24999 PR tree-optimization/77498
25000 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
25001 to non-constants over backedges.
25003 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
25005 PR rtl-optimization/80233
25006 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
25007 as last_combined_insn. Do not test for BARRIER_P separately.
25009 2017-03-29 Andreas Schwab <schwab@suse.de>
25012 * calls.c (prepare_call_address): Convert funexp to Pmode before
25013 copying to temp reg.
25015 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25017 PR tree-optimization/80158
25018 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
25019 Handle possible future case of more than one alternate
25021 (replace_rhs_if_not_dup): Likewise.
25022 (replace_one_candidate): Likewise.
25024 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
25026 PR rtl-optimization/80193
25027 * ira.c (ira): Do not check allocation for LRA.
25029 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
25031 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
25032 (nvptx_output_simt_exit): Declare.
25033 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
25034 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
25035 (init_softstack_frame): Move initialization of crtl->is_leaf to...
25036 (nvptx_declare_function_name): ...here. Emit declaration of local
25037 memory space buffer for omp_simt_enter insn.
25038 (nvptx_output_unisimt_switch): New.
25039 (nvptx_output_softstack_switch): New.
25040 (nvptx_output_simt_enter): New.
25041 (nvptx_output_simt_exit): New.
25042 * config/nvptx/nvptx.h (struct machine_function): New fields
25043 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
25044 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
25045 (UNSPECV_SIMT_EXIT): Ditto.
25046 (omp_simt_enter_insn): New insn.
25047 (omp_simt_enter): New expansion.
25048 (omp_simt_exit): New insn.
25049 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
25051 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
25052 (expand_GOMP_SIMT_ENTER_ALLOC): New.
25053 (expand_GOMP_SIMT_EXIT): New.
25054 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
25055 (GOMP_SIMT_ENTER_ALLOC): Ditto.
25056 (GOMP_SIMT_EXIT): Ditto.
25057 * target-insns.def (omp_simt_enter): New insn.
25058 (omp_simt_exit): Ditto.
25059 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
25061 (lower_rec_simd_input_clauses): Implement SIMT privatization.
25062 (lower_rec_input_clauses): Likewise.
25063 (lower_lastprivate_clauses): Handle SIMT privatization.
25065 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
25066 (ompdevlow_adjust_simt_enter): New.
25067 (find_simtpriv_var_op): New.
25068 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
25069 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
25071 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
25072 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
25073 (copy_decl_for_dup_finish): Ditto.
25075 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
25077 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
25080 * config/i386/i386.c (ix86_option_override_internal): Always
25081 allow -mpreferred-stack-boundary=3 for 64-bit targets.
25083 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25085 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
25087 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25089 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
25090 mark new edge's irreducible flag accordign to it.
25091 (vect_do_peeling): Check loop preheader edge's irreducible flag
25092 and pass it to function slpeel_add_loop_guard.
25094 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
25096 PR tree-optimization/80218
25097 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25098 Update block frequencies and counts.
25100 2017-03-28 Richard Biener <rguenther@suse.de>
25102 PR tree-optimization/78644
25103 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
25104 of a simplification result we may not use it at all.
25106 2017-03-28 Richard Biener <rguenther@suse.de>
25109 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
25110 without arguments, generate default definition of a SSA name.
25112 2017-03-28 Richard Biener <rguenther@suse.de>
25114 PR middle-end/80222
25115 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
25116 TYPE_REF_CAN_ALIAS_ALL references.
25117 * fold-const.c (fold_indirect_ref_1): Likewise.
25119 2017-03-28 Martin Liska <mliska@suse.cz>
25122 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
25123 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
25125 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
25126 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
25128 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
25129 (EXTRA_SPECS): Define.
25130 (SUBTARGET_EXTRA_SPECS): Likewise.
25131 (SUBTARGET_CPP_SPEC): Likewise.
25132 * config/arc/elf.h (EXTRA_SPECS): Renamed to
25133 SUBTARGET_EXTRA_SPECS.
25134 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
25136 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
25138 * config/arc/simdext.md (vst64_insn): Update pattern.
25139 (vld32wh_insn): Likewise.
25140 (vld32wl_insn): Likewise.
25141 (vld64_insn): Likewise.
25142 (vld32_insn): Likewise.
25144 2017-03-28 Marek Polacek <polacek@redhat.com>
25147 * fold-const.c (fold_comparison): Use protected_set_expr_location
25148 instead of SET_EXPR_LOCATION.
25150 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
25152 * tree.c (add_expr): Avoid name lookup warning.
25154 2017-03-27 Jeff Law <law@redhat.com>
25156 PR tree-optimization/80216
25157 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
25158 function name. Limit recursion depth.
25159 (record_temporary_equivalences): Corresponding changes.
25161 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
25163 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
25166 2017-03-27 Jakub Jelinek <jakub@redhat.com>
25169 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
25171 * cfgcleanup.c (reg_note_cfa_p): New array.
25172 (insns_have_identical_cfa_notes): New function.
25173 (old_insns_match_p): Don't cross-jump in between /f
25174 and non-/f instructions. If both i1 and i2 are frame related,
25175 verify all CFA notes, their order and content.
25177 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
25180 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
25181 HImode and SImode with zero extend to DImode to one insn.
25182 (bswap<mode>2_extenddi): Likewise.
25183 (bswapsi2_extenddi): Likewise.
25184 (bswaphi2_extendsi): Likewise.
25185 (bswaphi2): Combine bswap HImode and SImode into one insn.
25186 Separate memory insns from swapping register.
25187 (bswapsi2): Likewise.
25188 (bswap<mode>2): Likewise.
25189 (bswaphi2_internal): Delete, no longer used.
25190 (bswapsi2_internal): Likewise.
25191 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
25192 store, and gpr<-gpr swap insns.
25193 (bswap<mode>2_store): Likewise.
25194 (bswaphi2_reg): Register only splitter, combine with the splitter.
25195 (bswaphi2 splitter): Likewise.
25196 (bswapsi2_reg): Likewise.
25197 (bswapsi2 splitter): Likewise.
25198 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
25199 the insns into load, store, and register/register insns.
25200 (bswapdi2_ldbrx): Likewise.
25201 (bswapdi2_load): Likewise.
25202 (bswapdi2_store): Likewise.
25203 (bswapdi2_reg): Likewise.
25205 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
25207 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
25208 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
25210 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
25213 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
25215 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25216 special handling for target option conflicts between dform
25217 options (-mpower9-dform, -mpower9-dform-vector,
25218 -mpower9-dform-scalar) and -mno-direct-move.
25220 2017-03-27 Richard Biener <rguenther@suse.de>
25222 PR tree-optimization/80181
25223 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
25225 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25227 * config/arc/predicates.md (move_double_src_operand): Replace the
25228 call to move_double_src_operand with a call to address_operand.
25230 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25232 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
25233 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
25234 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
25236 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25238 * config/arc/predicates.md (long_immediate_loadstore_operand):
25239 Consider scaled addresses cases.
25241 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25243 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
25244 restored when in interrupt.
25245 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
25246 doesn't have delay slot.
25248 2017-03-27 Richard Biener <rguenther@suse.de>
25251 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
25252 inlined thunk clones.
25254 2017-03-27 Jakub Jelinek <jakub@redhat.com>
25257 * asan.c (instrument_derefs): Copy over last operand from
25258 original COMPONENT_REF to the new COMPONENT_REF with
25259 DECL_BIT_FIELD_REPRESENTATIVE.
25260 * ubsan.c (instrument_object_size): Likewise.
25262 2017-03-27 Richard Biener <rguenther@suse.de>
25264 PR tree-optimization/80170
25265 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
25266 sure DR/SCEV didnt fold in constants we do not see when looking
25267 at the reference base alignment.
25269 2017-03-27 Richard Biener <rguenther@suse.de>
25271 PR middle-end/80171
25272 * gimple-fold.c (fold_ctor_reference): Properly guard against
25273 NULL return value from canonicalize_constructor_val.
25275 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
25278 * config/i386/i386.c (ix86_expand_builtin)
25279 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
25280 flags reg setting and flags reg using instructions.
25281 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
25282 clobbering instructions to zero extend op2.
25284 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
25286 * doc/install.texi (Configuration) <--with-aix-soname>:
25287 Update link to AIX ld.
25289 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
25291 PR rtl-optimization/80160
25292 PR rtl-optimization/80159
25293 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
25294 reg_alternate_class into account.
25296 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
25299 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
25300 to consider in curr_insn_transform.
25302 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25304 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
25305 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
25306 and emit_mode_inner.
25308 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25310 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
25311 argument to the overloaded builtin variants. Use the new flag to
25312 deprecate certain builtin variants.
25313 * config/s390/s390-builtin-types.def: Add new builtin types.
25314 * config/s390/s390-builtins.h: Support new flags field for
25315 overloaded builtins.
25316 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
25317 (s390_macro_to_expand): Enable vector float data type.
25318 (s390_cpu_cpp_builtins_internal): Indicate support of the new
25319 builtins by incrementing the __VEC__ version number.
25320 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
25322 (s390_resolve_overloaded_builtin): Emit error messages depending
25323 on the builtin flags.
25324 * config/s390/s390.c (s390_expand_builtin): Support additional
25325 flags argument. Change error message to match the messages
25326 emitted in s390-c.c.
25327 * config/s390/s390.md: New UNSPEC_* constants.
25328 (op_type): Add new instruction types.
25329 * config/s390/vecintrin.h: Add new builtins and test data class
25331 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
25332 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
25333 (VEC_INEXACT, VEC_NOINEXACT): New constants.
25334 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
25335 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
25336 ("vec_mergel<mode>"): V_HW -> VEC_HW.
25338 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
25339 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
25340 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
25341 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
25343 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
25344 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
25345 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
25346 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
25348 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
25349 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
25350 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
25351 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
25352 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
25353 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
25354 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
25356 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
25357 ("vec_scatter_element<V_HW_4:mode>_DI")
25358 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
25359 ("vec_fpint<mode>", "vflls")
25360 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
25361 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
25362 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
25363 ("*vec_cmphe<mode>_cc"): ... these.
25365 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
25366 mode constant instead of magic value.
25368 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25370 * config/s390/s390.c (s390_expand_vec_compare): Support other
25371 vector floating point modes than just V2DF.
25372 (s390_expand_vcond): Likewise.
25373 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
25374 (s390_cannot_change_mode_class): Prevent mode changes between TF
25375 and V1TF in vector registers.
25376 * config/s390/s390.md (DF, SF): New mode attributes.
25377 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
25378 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
25379 SFmode support for VRs.
25380 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
25382 (VFT, VF_HW): New mode iterators.
25383 (vw, sdx): New mode attributes.
25384 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
25385 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
25386 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
25387 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
25388 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
25389 also the new vector floating point modes. Renaming to ...
25391 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
25392 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
25393 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
25394 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
25395 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
25396 ("vec_unordered<mode>"): ... these.
25398 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
25399 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
25400 ("*vec_extendv2df"): New insn definitions.
25402 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25404 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
25405 ("mulditi3_2", "*muldi3_sign"): New patterns.
25406 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
25407 rename the pattern definition.
25409 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25411 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
25413 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
25415 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25417 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
25418 instruction if possible.
25419 * config/s390/vector.md (vec_halfnumelts): New mode
25421 ("*vec_vllezlf<mode>"): New pattern.
25423 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25425 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
25426 ("popcountv4si2", "popcountv2di2"): Rename to ...
25427 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
25428 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
25430 ("popcount<mode>2_vxe"): New pattern.
25432 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25434 * common/config/s390/s390-common.c (processor_flags_table): Add
25436 * config.gcc: Add arch12.
25437 * config/s390/driver-native.c (s390_host_detect_local_cpu):
25438 Default to arch12 for unknown CPU model numbers.
25439 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
25440 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
25441 PROCESSOR_max sanity check.
25442 * config/s390/s390-opts.h (enum processor_type): Add
25444 * config/s390/s390.c (processor_table): Add arch12.
25445 (s390_expand_builtin): Add check for B_VXE flag.
25446 (s390_issue_rate): Add PROCESSOR_ARCH12.
25447 (s390_get_sched_attrmask): Likewise.
25448 (s390_get_unit_mask): Likewise.
25449 (s390_sched_score): Enable z13 scheduling for arch12.
25450 (s390_sched_reorder): Likewise.
25451 (s390_sched_variable_issue): Likewise.
25452 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
25454 (s390_tune_attr): Use z13 scheduling also for arch12.
25455 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
25456 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
25457 (TARGET_VXE_P): New macros.
25458 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
25459 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
25460 * config/s390/s390.opt: Add arch12 as processor_type.
25462 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25464 * config/s390/s390.md
25465 ("fixuns_truncdddi2", "fixuns_trunctddi2")
25466 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
25467 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
25469 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
25470 Rename expanders to ...
25472 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
25473 ("fixuns_truncdddi2_emu"): ... these.
25475 ("fixuns_trunc<mode>si2_emu"): New expander.
25477 ("*fixuns_truncdfdi2_z13"): Rename to ...
25478 ("*fixuns_truncdfdi2_vx"): ... this.
25480 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25482 * config/s390/2964.md: Remove the single element vector compare
25483 instructions which are no longer used.
25484 * config/s390/s390.c (s390_select_ccmode): Remove handling of
25486 (s390_canonicalize_comparison): Remove handling of DFmode
25488 (s390_expand_vec_compare_scalar): Remove function.
25489 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
25490 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
25492 ("*cmp<mode>_ccs"): Add wfcdb instruction.
25494 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25496 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
25498 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
25499 will anyway by matched by mov<mode>_64dfp.
25501 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25503 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
25504 vlef/vstef. Add missing operand to vleif.
25506 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25508 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
25509 pair for all vector types with 64 bit elements.
25510 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
25511 * config/s390/vector.md (V_HW_64): ... here.
25512 (V_128_NOSINGLE): New mode iterator.
25513 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
25514 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
25515 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
25516 ("*vec_load_pairv2di"): Change to ...
25517 ("*vec_load_pair<mode>"): ... this one.
25519 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25521 * config/s390/constraints.md: Add comments.
25522 (jKK): Reject element sizes > 8 bytes.
25523 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
25525 * config/s390/s390.md: Add the s_operand checks formerly in
25526 s390_split_ok_p to various splitters where they are still
25528 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
25529 for 128 bit vectors. Plus two splitters.
25531 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25533 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
25536 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25539 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
25540 error if the boundary argument is not constant.
25542 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25544 PR rtl-optimization/80112
25545 * loop-doloop.c (doloop_condition_get): Don't check condition
25546 if cmp isn't SET with IF_THEN_ELSE src.
25548 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25550 PR tree-optimization/80158
25551 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
25552 replacing a candidate statement, also replace it for the
25553 candidate's alternate interpretation.
25554 (replace_rhs_if_not_dup): Likewise.
25555 (replace_one_candidate): Likewise.
25557 2017-03-24 Richard Biener <rguenther@suse.de>
25559 PR tree-optimization/80167
25560 * graphite-isl-ast-to-gimple.c
25561 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
25563 (translate_isl_ast_to_gimple::get_rename): Likewise.
25565 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25567 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
25568 handling of certain combinations of target options, including the
25569 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
25570 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
25572 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25575 * config/arm/arm.md (*load_multiple): Add reload_completed to
25576 matching condition.
25578 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25579 Richard Biener <rguenth@suse.de>
25581 PR tree-optimization/79908
25582 PR tree-optimization/80136
25583 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25584 been cast away, gimplify_and_add suffices.
25586 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
25588 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
25590 2017-03-23 Richard Biener <rguenther@suse.de>
25592 PR tree-optimization/80032
25593 * gimplify.c (gimple_push_cleanup): Forced unconditional
25594 cleanups still have to go to the conditional_cleanups
25597 2017-03-22 Jakub Jelinek <jakub@redhat.com>
25599 PR tree-optimization/80072
25600 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
25602 (next_operand_entry_id): Change type to unsigned int.
25603 (sort_by_operand_rank): Make sure to return the right return value
25604 even if unsigned fields are bigger than INT_MAX.
25605 (struct oecount): Change cnt and id type to unsigned int.
25606 (oecount_hasher::equal): Formatting fix.
25607 (oecount_cmp): Make sure to return the right return value
25608 even if unsigned fields are bigger than INT_MAX.
25609 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
25612 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
25613 TREE_READONLY on result if writing it more than once.
25616 * doc/invoke.texi (-fsanitize=thread): Document that with
25617 -fnon-call-exceptions atomics are not able to throw
25621 * tsan.c: Include tree-eh.h.
25622 (instrument_builtin_call): Call maybe_clean_eh_stmt or
25623 maybe_clean_or_replace_eh_stmt where needed.
25624 (instrument_memory_accesses): Add cfg_changed argument.
25625 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
25626 if it returned true.
25627 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
25629 PR rtl-optimization/63191
25630 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
25631 wrapper function, moved the whole old content into ...
25632 (ix86_delegitimize_address_1): ... this. New inline function.
25633 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
25634 true as last argument instead of ix86_delegitimize_address.
25636 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25638 * config/aarch64/aarch64.c (generic_branch_cost): Copy
25639 cortexa57_branch_cost.
25641 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25643 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
25645 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25648 * doc/md.texi (Constraints): Document wA constraint.
25649 * config/rs6000/constraints.md (wA): New.
25650 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
25651 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
25652 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
25653 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
25655 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
25658 * gimplify.c (is_oacc_declared): New function.
25659 (oacc_default_clause): Use it to set default flags for acc declared
25660 variables inside parallel regions.
25661 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
25662 declared variables.
25663 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
25664 declare attribute to any decl as necessary.
25666 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
25669 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
25670 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
25671 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
25672 (arm_arch_lpae): This.
25673 * config/arm/arm.c (arm_arch7ve): Rename into ...
25674 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
25675 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
25678 2017-03-22 Martin Liska <mliska@suse.cz>
25681 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
25682 error message instead of an ICE.
25684 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25686 * doc/extend.texi (6.11 Additional Floating Types): Revise.
25688 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
25690 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25692 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25695 2017-03-21 Martin Sebor <msebor@redhat.com>
25697 * doc/extend.texi: Use "cannot" instead of "can't."
25698 * doc/hostconfig.texi: Same.
25699 * doc/install.texi: Same.
25700 * doc/invoke.texi: Same.
25701 * doc/loop.texi: Same.
25702 * doc/md.texi: Same.
25703 * doc/objc.texi: Same.
25704 * doc/rtl.texi: Same.
25705 * doc/tm.texi: Same.
25706 * doc/tm.texi.in: Same.
25707 * doc/trouble.texi: Same.
25709 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
25712 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
25713 (collect_checksum_attributes): Set it.
25714 (die_checksum_ordered): Use it.
25716 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25718 PR tree-optimization/79908
25719 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
25720 change: For a VA_ARG whose LHS has been cast away, use
25721 force_gimple_operand to construct the side effects.
25723 2017-03-21 David Malcolm <dmalcolm@redhat.com>
25725 PR translation/80001
25726 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
25727 more amenable to translation.
25728 (oacc_loop_auto_partitions): Likewise.
25730 2017-03-21 Marek Polacek <polacek@redhat.com>
25731 Martin Sebor <msebor@redhat.com>
25733 PR tree-optimization/80109
25734 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
25735 on INTEGRAL_TYPE_P.
25737 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25738 Segher Boessenkool <segher@kernel.crashing.org>
25741 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
25742 check reg_used_between_p between insn and one of succ or succ2
25743 depending on if succ is artificial insn not inserted into insn
25746 2017-03-21 Martin Liska <mliska@suse.cz>
25748 PR gcov-profile/80081
25749 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
25750 * doc/gcc.texi: Include gcov-dump stuff.
25751 * doc/gcov-dump.texi: New file.
25753 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
25755 PR rtl-optimization/79150
25756 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
25757 conditional jump, if the jump is the last insn of the loop.
25759 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25760 Richard Biener <rguenth@suse.de>
25762 PR tree-optimization/79908
25763 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25764 been cast away, use force_gimple_operand to construct the side
25767 2017-03-21 Martin Liska <mliska@suse.cz>
25769 PR libfortran/79956
25770 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
25773 2017-03-21 Brad Spengler <spender@grsecurity.net>
25776 * plugin.c (htab_hash_plugin): New function.
25777 (add_new_plugin): Use it and adjust.
25778 (parse_plugin_arg_opt): Adjust.
25779 (init_one_plugin): Likewise.
25781 2017-03-21 Richard Biener <rguenther@suse.de>
25783 PR tree-optimization/80032
25784 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
25785 if set force the cleanup to happen unconditionally.
25786 (gimplify_target_expr): Push inserted clobbers with force_uncond
25787 to avoid them being removed by control-dependent DCE.
25789 2017-03-21 Richard Biener <rguenther@suse.de>
25791 PR tree-optimization/80122
25792 * tree-inline.c (copy_bb): Do not expans va-arg packs or
25793 va_arg_pack_len when the inlined call stmt requires pack
25795 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
25797 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25800 * tsan.c (instrument_builtin_call): If the memory model argument
25801 is not a constant, assume it is valid.
25804 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
25807 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25809 PR rtl-optimization/79910
25810 * combine.c (can_combine_p): Do not allow combining an I0 or I1
25811 if its dest is used by an insn before I2 (other than the combined
25812 insns themselves, which are properly handled already).
25814 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25817 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25819 * combine.c (record_used_regs): New static function.
25820 (try_combine): Handle situations where there is an additional
25821 instruction between I2 and I3 which needs to have a LOG_LINK
25825 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25827 * combine.c (try_combine): Delete redundant i1 test. Call
25828 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25830 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25833 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
25834 alternatives 13/14.
25836 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25838 PR tree-optimization/80054
25839 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
25840 the optimization if a PHI or any of its arguments is not dominated
25841 by the candidate's basis. Use gphi* rather than gimple* as
25843 (replace_profitable_candidates): Clean up a gimple* variable that
25844 should be a gphi* variable.
25846 2017-03-20 Martin Sebor <msebor@redhat.com>
25849 * doc/extend.texi (attribute constructor): Document present limitation.
25851 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
25854 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
25855 __POWER9_VECTOR__ #ifdef control, change template definition to
25856 use Power9-specific built-in function.
25857 (vec_any_eq): Likewise.
25858 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
25859 to control outcomes from this test.
25860 (vector_ae_<mode>p): For VEC_F modes, likewise.
25862 2017-03-20 Ian Lance Taylor <iant@google.com>
25864 * config/i386/i386.c (ix86_function_regparm): Save an extra
25865 register for -fsplit-stack with DECL_STATIC_CHAIN.
25867 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25870 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
25871 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
25873 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25875 * config/riscv/riscv.c (riscv_print_operand): Use "fence
25877 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
25880 2017-03-20 Marek Polacek <polacek@redhat.com>
25883 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
25885 2017-03-20 Richard Biener <rguenther@suse.de>
25887 PR tree-optimization/80113
25888 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
25889 allocate extra SSA name for PHI def.
25890 (add_close_phis_to_outer_loops): Likewise.
25891 (add_close_phis_to_merge_points): Likewise.
25892 (copy_loop_close_phi_args): Likewise.
25893 (copy_cond_phi_nodes): Likewise.
25895 2017-03-20 Martin Liska <mliska@suse.cz>
25897 PR middle-end/79753
25898 * tree-chkp.c (chkp_build_returned_bound): Do not build
25899 returned bounds for a LHS that's not a BOUNDED_P type.
25901 2017-03-20 Martin Liska <mliska@suse.cz>
25905 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
25906 COMPLEX_CST and VECTOR_CST.
25908 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25911 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
25912 target operand. A new splitter adds the clobber statement in case
25913 the target operand is dead anyway.
25915 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
25917 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
25918 to age-old versions of binutils and glibc.
25920 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
25922 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
25924 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
25926 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
25928 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
25930 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
25931 requirement for binutils 2.13.
25933 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25935 * combine.c (try_combine): Delete redundant i1 test. Call
25936 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25938 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
25940 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
25941 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
25943 <riscv64-*-elf>: Re-arrange section
25944 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
25945 <riscv32-*-linux>: Likewise.
25946 <riscv64-*-elf>: Likewise
25947 <riscv64-*-linux>: Likewise.
25949 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
25952 * aarch64.opt(verbose-cost-dump): Fix typo.
25954 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
25957 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
25958 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
25960 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25962 * reload.c (find_reloads): When reloading a nonoffsettable address,
25963 use RELOAD_OTHER for it and its address reloads.
25965 PR rtl-optimization/79910
25966 * combine.c (record_used_regs): New static function.
25967 (try_combine): Handle situations where there is an additional
25968 instruction between I2 and I3 which needs to have a LOG_LINK
25971 2017-03-17 Jeff Law <law@redhat.com>
25973 PR tree-optimization/71437
25974 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
25975 conditional in the hash table first.
25976 (vrp_dom_walker::before_dom_children): Extract condition from
25977 ASSERT_EXPR. Record condition, its inverion and any implied
25978 conditions as well.
25980 2017-03-17 Marek Polacek <polacek@redhat.com>
25981 Markus Trippelsdorf <markus@trippelsdorf.de>
25983 PR tree-optimization/80079
25984 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
25987 2017-03-17 Richard Biener <rguenther@suse.de>
25989 PR middle-end/80075
25990 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
25991 Properly verify the LHS before the RHS possibly claims to be
25993 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
25996 2017-03-17 Martin Jambor <mjambor@suse.cz>
25998 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
25999 (List of -O2 options): Likewise.
26000 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
26003 2017-03-17 Tom de Vries <tom@codesourcery.com>
26005 * gcov-dump.c (print_usage): Print bug_report_url.
26007 2017-03-17 Richard Biener <rguenther@suse.de>
26009 PR middle-end/80050
26010 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
26011 (parser::peek): Likewise.
26013 2017-03-17 Richard Biener <rguenther@suse.de>
26015 PR tree-optimization/80048
26016 * sese.c (free_sese_info): Properly release rename_map and
26017 copied_bb_map elements.
26019 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
26021 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
26022 Add linked-list forward and backlinks. Insert on
26023 construction, remove on destruction.
26024 (class pass_store_merging): Add m_stores_head field.
26025 (pass_store_merging::terminate_and_process_all_chains):
26026 Iterate over m_stores_head list.
26027 (pass_store_merging::terminate_all_aliasing_chains):
26029 (pass_store_merging::execute): Check for debug stmts first.
26030 Push new chains onto the m_stores_head stack.
26032 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
26035 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
26036 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
26037 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
26039 2017-03-16 Jeff Law <law@redhat.com>
26041 PR tree-optimization/71437
26042 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
26043 member function. Implementation moved into after_dom_children
26044 member function and into the threader's thread_outgoing_edges
26046 (dom_opt_dom_walker::after_dom_children): Simplify by moving
26047 some code into new thread_outgoing_edges.
26048 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
26049 definition. Simplify marker handling (do it here). Assume we always
26050 have the available expression and the const/copies tables.
26051 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
26053 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
26054 * tree-vrp.c (equiv_stack): No longer file scoped.
26055 (vrp_dom_walker): New class.
26056 (vrp_dom_walker::before_dom_children): New member function.
26057 (vrp_dom_walker::after_dom_children): Likewise.
26058 (identify_jump_threads): Setup domwalker. Use it rather than
26059 walking edges in a random order by hand. Simplify setup/finalization.
26060 (finalize_jump_threads): Remove.
26061 (vrp_finalize): Do not call identify_jump_threads here.
26062 (execute_vrp): Do it here instead and call thread_through_all_blocks
26065 PR tree-optimization/71437
26066 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
26068 (simplify_stmt_for_jump_threading): Add basic_block argument. All
26070 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
26071 (dom_opt_dom_walker::thread_across_edge): Remove
26072 handle_dominating_asserts argument. All callers changed.
26073 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
26074 changes. Remove calls to lhs_of_dominating_assert. Other
26075 uses of handle_dominating_asserts turn into unconditional code
26076 (simplify_control_stmt_condition_1): Likewise.
26077 (simplify_control_stmt_condition): Likewise.
26078 (thread_through_normal_block, thread_across_edge): Likewise.
26079 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
26080 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
26081 object if it is not an SSA_NAME.
26082 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
26083 before calling into the VRP specific simplifiers.
26084 (identify_jump_threads): Remove handle_dominating_asserts
26087 2017-03-16 Jakub Jelinek <jakub@redhat.com>
26090 * tree-diagnostic.c (default_tree_printer): No longer static.
26091 * tree-diagnostic.h (default_tree_printer): New prototype.
26093 2017-03-16 Tamar Christina <tamar.christina@arm.com>
26095 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
26096 Change ins into fmov.
26098 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26100 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
26101 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
26102 Use h_con constraint for operand 1.
26103 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
26104 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
26106 2017-03-15 Jeff Law <law@redhat.com>
26108 PR tree-optimization/71437
26109 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
26110 (record_temporary_equivalences): Use it.
26112 PR tree-optimization/71437
26113 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
26114 tree-ssa-scopedtables.
26115 (lookup_avail_expr, build_and_record_new_cond): Likewise.
26116 (record_conditions, record_cond, vuse_eq): Likewise.
26117 (record_edge_info): Adjust to API tweak of record_conditions.
26118 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
26119 (record_temporary_equivalences, optimize_stmt): Likewise.
26120 (eliminate_redundant_computations): Likewise.
26121 (record_equivalences_from_stmt): Likewise.
26122 * tree-ssa-scopedtables.c: Include options.h and params.h.
26123 (vuse_eq): New function, moved from tree-ssa-dom.c
26124 (build_and_record_new_cond): Likewise.
26125 (record_conditions): Likewise. Accept vector of conditions rather
26126 than edge_equivalence structure for first argument.
26127 for the first argument.
26128 (avail_exprs_stack::lookup_avail_expr): New member function, moved
26129 from tree-ssa-dom.c.
26130 (avail_exprs_stack::record_cond): Likewise.
26131 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
26132 from tree-ssa-dom.c.
26133 (avail_exprs_stack): Add new member functions lookup_avail_expr
26135 (record_conditions): Declare.
26137 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
26140 * lra-constraints.c (process_alt_operands): Increase reject for
26141 reloading an input/output operand.
26143 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
26146 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
26147 insns to convert from signed/unsigned char/short to IEEE 128-bit
26149 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
26151 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
26154 * config/i386/i386.c (ix86_vector_duplicate_value): Create
26155 subreg of inner mode for values already in registers.
26157 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
26159 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
26160 iteration reg is used after the loop.
26162 2017-03-14 Martin Sebor <msebor@redhat.com>
26164 PR tree-optimization/79800
26165 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
26166 precision in negative-positive range.
26167 (format_floating): Call non-const overload with adjusted precision.
26169 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26172 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
26175 2017-03-14 Martin Sebor <msebor@redhat.com>
26177 PR middle-end/80020
26178 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
26179 * builtins.def (aligned_alloc): Use it.
26182 * Makefile.in (GTFILES): Add calls.c.
26183 * calls.c: Include "gt-calls.h".
26185 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
26187 PR rtl-optimization/79728
26188 * regs.h (struct target_regs): New field
26189 x_contains_allocatable_regs_of_mode.
26190 (contains_allocatable_regs_of_mode): New macro.
26191 * reginfo.c (init_reg_sets_1): Initialize it, and change
26192 contains_reg_of_mode so it includes global regs as well.
26193 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
26194 rather than contains_regs_of_mode.
26196 2017-03-14 Martin Liska <mliska@suse.cz>
26198 * doc/invoke.texi: Document options that can't be combined with
26199 -fcheck-pointer-bounds.
26201 2017-03-14 Martin Liska <mliska@suse.cz>
26203 PR middle-end/79831
26204 * doc/invoke.texi (-Wchkp): Document the option.
26206 2017-03-14 Martin Liska <mliska@suse.cz>
26208 * Makefile.in: Install gcov-dump.
26210 2017-03-14 Martin Liska <mliska@suse.cz>
26212 * multiple_target.c (expand_target_clones): Bail out for
26213 an invalid attribute.
26215 2017-03-14 Richard Biener <rguenther@suse.de>
26217 * alias.c (struct alias_set_entry): Pack properly.
26218 * cfgloop.h (struct loop): Likewise.
26219 * cse.c (struct set): Likewise.
26220 * ipa-utils.c (struct searchc_env): Likewise.
26221 * loop-invariant.c (struct invariant): Likewise.
26222 * lra-remat.c (struct cand): Likewise.
26223 * recog.c (struct change_t): Likewise.
26224 * rtl.h (struct address_info): Likewise.
26225 * symbol-summary.h (function_summary): Likewise.
26226 * tree-loop-distribution.c (struct partition): Likewise.
26227 * tree-object-size.c (struct object_size_info): Likewise.
26228 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
26229 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
26230 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
26231 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
26232 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
26233 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
26234 (struct _stmt_vec_info): Likewise.
26236 2017-03-14 Martin Liska <mliska@suse.cz>
26239 * multiple_target.c (create_dispatcher_calls): Check that
26240 a target can create a function dispatcher.
26242 2017-03-14 Martin Liska <mliska@suse.cz>
26245 * multiple_target.c (expand_target_clones): Drop local.local
26246 flag for default implementation.
26248 2017-03-14 Richard Biener <rguenther@suse.de>
26250 PR tree-optimization/80030
26251 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
26253 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
26255 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
26256 gcc_fallthrough() instead of __attribute__((fallthrough));
26258 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
26260 * doc/gcc.texi: Remove "up" link to (DIR).
26261 * doc/gccint.texi: Ditto.
26263 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
26265 * doc/install.texi (Specific) <avr>: Remove reference to
26268 2017-03-13 Jeff Law <law@redhat.com>
26270 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
26271 attribute rather than comments.
26273 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
26274 match_scratch operand is highest.
26276 2017-03-13 Martin Liska <mliska@suse.cz>
26278 PR middle-end/78339
26279 * ipa-pure-const.c (warn_function_noreturn): If the declarations
26280 is a CHKP clone, use original declaration.
26282 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26284 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
26285 (arc_conditional_register_usage): Use a different allocation order
26286 when optimizing for size.
26287 * common/config/arc/arc-common.c (arc_option_optimization_table):
26288 Section anchors default on when optimizing for size.
26290 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26292 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
26294 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26296 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
26297 * config/arc/arc.md (cpu_facility): Add cd variant.
26298 (*movqi_insn): Add code density variant.
26299 (*movhi_insn): Likewise.
26300 (*movqi_insn): Likewise.
26301 (*addsi3_mixed): Likewise.
26302 (subsi3_insn): Likewise.
26304 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26306 * config/arc/arc.md (movsi_cond_exec): Update constraint.
26308 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26310 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
26311 expressions with MINUS and UNARY ops.
26313 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26316 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
26318 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
26319 between vec_select and vector argument.
26320 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
26321 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
26322 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
26323 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
26324 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
26325 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
26327 2017-03-13 Richard Biener <rguenther@suse.de>
26330 * params.def (vect-max-peeling-for-alignment): Fix typo.
26332 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26334 * doc/install.texi (Specific) <mips-*-*>: Remove description of
26335 issue that only occurred with binutils below 2.18.
26337 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26339 * doc/install.texi (Specific) <cris-axis-elf>: No longer
26340 refer to binutils 2.11/2.12 minimum.
26342 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26344 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
26345 ftp.kernel.org and simplify binutils requirement.
26347 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
26349 * doc/invoke.texi (Warning Options): Fix spelling of link-time
26351 (Optimize Options): Ditto. Also remove redundancy.
26353 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26355 PR translation/79848
26356 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
26358 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
26359 to G_ to avoid double translation.
26361 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26363 PR translation/79923
26364 * auto-profile.c (get_combined_location): Convert leading
26365 character of diagnostics to lower case and remove trailing period.
26366 (read_profile): Likewise for various diagnostics.
26367 * config/arm/arm.c (arm_option_override): Remove trailing period
26368 from various diagnostics.
26369 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
26370 (msp430_expand_delay_cycles): Likewise.
26372 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26375 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
26376 full command-line argument, rather than just "str".
26377 (aarch64_validate_march): Likewise.
26378 (aarch64_validate_mtune): Likewise.
26380 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
26382 PR rtl-optimization/78911
26383 * lra-assigns.c (must_not_spill_p): New function.
26384 (spill_for): Use it.
26386 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26388 PR tree-optimization/79981
26389 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
26390 ATOMIC_COMPARE_EXCHANGE ifn result.
26391 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
26392 IFN_ATOMIC_COMPARE_EXCHANGE.
26394 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26397 * opts.c (parse_sanitizer_options): Add missing question mark to
26398 "did you mean" message.
26400 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26402 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
26404 (VMULEUH_UNS): Likewise.
26405 (VMULOUB_UNS): Likewise.
26406 (VMULOUH_UNS): Likewise.
26407 * config/rs6000/rs6000.c (builtin_function_type): Remove
26408 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
26410 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26413 * read-rtl-function.c (function_reader::read_rtx_operand): Update
26414 x with result of extra_parsing_for_operand_code_0.
26415 (function_reader::extra_parsing_for_operand_code_0): Convert
26416 return type from void to rtx, returning x. When reading
26417 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
26418 larger size containing struct block_symbol.
26420 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
26422 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
26423 -mfloat128-hardware without -m64.
26425 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
26428 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
26429 entries to the case statement that marks unsigned arguments to
26430 overloaded functions.
26432 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26434 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
26435 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
26437 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
26440 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
26441 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
26443 2017-03-10 Martin Liska <mliska@suse.cz>
26447 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
26448 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
26451 2017-03-10 Olivier Hainque <hainque@adacore.com>
26453 * tree-switch-conversion (array_value_type): Start by resetting
26454 candidate type to it's main variant.
26456 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26458 PR rtl-optimization/79909
26459 * combine.c (try_combine): Use simplify_replace_rtx on individual
26460 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
26461 of the whole CALL_INSN_FUNCTION_USAGE.
26463 PR tree-optimization/79972
26464 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
26465 get_range_info on SSA_NAMEs. Formatting fixes.
26467 2017-03-10 Richard Biener <rguenther@suse.de>
26468 Jakub Jelinek <jakub@redhat.com>
26470 PR tree-optimization/77975
26471 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
26472 edge to be constant.
26473 (get_val_for): For constant x return it. Formatting fix.
26474 (loop_niter_by_eval): Avoid pointless looping if the next iteration
26475 would use the same bases as the current one.
26477 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26479 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
26480 instead of vec_select for V1TImode.
26481 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
26483 (VSX_LE_128): Add V1TI to this mode iterator.
26484 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
26485 (*vsx_le_perm_store_<mode>): Likewise.
26486 (pre-reload splitter for VSX stores): Likewise.
26487 (post-reload splitter for VSX stores): Likewise.
26488 (*vsx_xxpermdi2_le_<mode>): Likewise.
26489 (*vsx_lxvd2x2_le_<mode>): Likewise.
26490 (*vsx_stxvd2x2_le_<mode>): Likewise.
26492 2017-03-09 Michael Eager <eager@eagercon.com>
26494 Correct failures with --enable-checking=yes,rtl.
26496 * config/microblaze/microblaze.c (microblaze_expand_shift):
26497 Replace GET_CODE test with CONST_INT_P and INTVAL test with
26498 test for const0_rtx.
26499 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
26500 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
26502 2017-03-09 Richard Biener <rguenther@suse.de>
26504 PR tree-optimization/79977
26505 * graphite-scop-detection.c (scop_detection::merge_sese):
26506 Handle the case of extra exits to blocks dominating the entry.
26508 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
26510 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26513 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
26515 PR rtl-optimization/79949
26516 * lra-constraints.c (process_alt_operands): Check memory when
26517 trying to predict a cycle. Print about the overall increase.
26519 2017-03-09 Richard Biener <rguenther@suse.de>
26521 PR middle-end/79971
26522 * gimple-expr.c (useless_type_conversion_p): Preserve
26523 TYPE_SATURATING for fixed-point types.
26525 2017-03-09 Richard Biener <rguenther@suse.de>
26528 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
26529 alignment of BLKmode params.
26531 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26534 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
26535 (VALL_NO_V2Q): Likewise.
26537 * config/aarch64/aarch64-simd.md
26538 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
26540 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
26541 VALL_NO_V2Q mode iterator.
26542 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
26544 2017-03-09 Martin Liska <mliska@suse.cz>
26546 PR tree-optimization/79631
26547 * tree-chkp-opt.c (chkp_is_constant_addr): Call
26548 tree_int_cst_sign_bit just for INTEGER constants.
26550 2017-03-09 Martin Liska <mliska@suse.cz>
26554 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
26557 2017-03-09 Marek Polacek <polacek@redhat.com>
26560 * tree.c (inchash::add_expr): Handle TREE_VEC.
26562 2017-03-09 Martin Liska <mliska@suse.cz>
26565 (chkp_narrow_size_and_offset): New function.
26566 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
26567 (void chkp_parse_bit_field_ref): New function.
26568 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
26569 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
26571 2017-03-09 Martin Liska <mliska@suse.cz>
26574 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
26575 (chkp_find_bounds_1): Remove gcc_unreachable.
26577 2017-03-09 Jakub Jelinek <jakub@redhat.com>
26580 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
26581 BUILT_IN_SYNC*, determine the access type from the size suffix and
26582 always build a MEM_REF with that type. Handle forgotten
26583 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
26586 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
26587 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
26588 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
26589 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
26590 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
26591 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
26592 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
26593 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
26594 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
26595 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
26596 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
26597 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
26598 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
26599 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
26600 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
26601 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
26602 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
26603 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
26604 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
26605 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
26606 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
26607 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
26608 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
26609 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
26610 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
26611 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
26612 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
26613 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
26614 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
26615 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
26616 definitions outside of __OPTIMIZE__ guarded section.
26619 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
26620 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
26621 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
26622 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
26625 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26627 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
26628 ("vfenez<mode>"): Add missing constraints.
26630 2017-03-08 Martin Sebor <msebor@redhat.com>
26633 * config/nds32/nds32.c (nds32_option_override):
26634 Fix misspelled diagnostic.
26636 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26639 * gimplify.c (gimplify_omp_for): Replace index var in outer
26640 taskloop statement with an artificial variable and add
26641 OMP_CLAUSE_PRIVATE clause for it.
26643 2017-03-08 Richard Biener <rguenther@suse.de>
26645 PR tree-optimization/79955
26646 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
26647 for accesses that are completely outside of the variable.
26649 2017-03-08 Andrew Haley <aph@redhat.com>
26651 PR tree-optimization/79943
26652 * tree-ssa-loop-split.c (compute_new_first_bound): When
26653 calculating the new upper bound, (END-BEG) should be added, not
26656 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26658 * config/avr/avr.md (setmemhi): Make sure match_dup
26659 operand number comes before match_scratch.
26661 2017-03-08 Richard Biener <rguenther@suse.de>
26663 PR tree-optimization/79920
26664 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
26665 with ncopies == 1 to ...
26666 (vect_transform_slp_perm_load): ... here. Properly compute
26667 all element loads by iterating VF times over the group. Do
26668 not handle ncopies (computed in a broken way) in
26669 vect_create_mask_and_perm.
26671 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26674 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
26675 is a uniform vector, use uniform_vector_p return value instead of
26676 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
26678 2017-03-07 Marek Polacek <polacek@redhat.com>
26680 PR middle-end/79809
26681 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
26682 (alloca_call_type): Likewise.
26684 2017-03-07 Martin Liska <mliska@suse.cz>
26686 * gcov.c (process_args): Put comment to correct location.
26688 2017-03-07 Martin Liska <mliska@suse.cz>
26690 PR middle-end/68270
26691 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
26692 Use array_at_struct_end_p instead of DECL_CHAIN (field).
26693 (chkp_narrow_bounds_for_field): Likewise.
26694 (chkp_parse_array_and_component_ref): Pass one more argument to
26697 2017-03-07 Richard Biener <rguenther@suse.de>
26699 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
26702 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
26704 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
26705 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
26707 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26710 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
26711 to end of description.
26712 (PARAM_MAX_STORES_TO_MERGE): Likewise.
26714 2017-03-07 Jakub Jelinek <jakub@redhat.com>
26716 PR rtl-optimization/79901
26717 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
26719 (*avx512f_<code><mode>3<mask_name>): ... this.
26720 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
26721 iterator instead of VI8_AVX2_AVX512BW.
26723 PR rtl-optimization/79901
26724 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
26725 min/max expander, expand it using expand_vec_cond_expr.
26728 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
26731 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26734 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
26735 to void * for PCH reasons.
26736 * dwarf2out.c (output_loc_operands, output_die): Cast
26737 v.val_vec.array to unsigned char *.
26739 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
26742 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
26745 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
26747 PR rtl-optimization/79571
26748 * lra-constraints.c (process_alt_operands): Calculate static
26749 reject and subtract it from overall when only addresses will be
26752 2017-03-06 Julia Koval <julia.koval@intel.com>
26755 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
26756 incoming stack boundary to 128 for 64-bit targets.
26758 2017-03-06 Richard Biener <rguenther@suse.de>
26760 PR tree-optimization/79894
26761 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
26762 to NULL after folding it.
26764 2017-03-06 Richard Biener <rguenther@suse.de>
26766 PR tree-optimization/79824
26767 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
26768 check disabling peeling for gaps.
26770 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
26772 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26773 attributes): Document gettimeofday.
26775 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26777 * config/s390/s390.c (s390_option_override_internal): Set
26778 PARAM_MIN_VECT_LOOP_BOUND
26780 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26782 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
26783 * config/s390/s390.md: Likewise.
26785 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26788 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
26789 (<avx2_avx512>_perm<mode>): Rename to ...
26790 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
26792 (<avx512>_perm<mode>_mask): Rename to ...
26793 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
26795 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
26796 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
26797 instead of VI8F_256_512.
26798 (avx512f_perm<mode>): New define_expand.
26799 (avx512f_perm<mode>_mask): Likewise.
26800 (avx512f_perm<mode>_1<mask_name>): New define_insn.
26801 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
26803 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26805 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
26806 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
26808 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
26810 2017-03-06 Martin Liska <mliska@suse.cz>
26813 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
26814 when having a SSA NAME w/o VAR_DECL assigned to it.
26816 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26818 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
26819 msa_dpsub_<su>_d): Fix MODE for vec_select.
26821 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26823 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
26825 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
26827 2017-03-06 Richard Biener <rguenther@suse.de>
26829 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
26830 * plugin.c (register_plugin_info): Likewise.
26831 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
26833 2017-03-05 Jakub Jelinek <jakub@redhat.com>
26835 * config/i386/sse.md (sse_storehps, sse_storelps,
26836 avx_<castmode><avxsizesuffix>_<castmode>,
26837 avx512f_<castmode><avxsizesuffix>_<castmode>,
26838 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
26839 in condition that at least one operand is not a MEM.
26841 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26843 PR middle-end/79805
26844 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
26845 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
26847 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
26848 gimple_call_nothrow_p flag based on whether original builtin can throw.
26849 If it can, emit following stmts on the fallthrough edge.
26850 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
26851 don't create new bb if inserting just debug stmts on the edge, try to
26852 insert them on the fallthru bb or just reset debug stmts.
26854 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
26857 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
26858 restore recog_data (including the operand rtxes inside it) around
26859 the call to get_insn_template.
26861 2017-03-03 Martin Sebor <msebor@redhat.com>
26863 PR tree-optimization/79699
26864 * context.c (context::~context): Free MPFR caches to avoid
26865 a memory leak on program exit.
26867 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26869 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
26870 Use wide_int::ulow () instead of .elt (0).
26872 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26874 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
26875 (*pushxf): Limit oF constraint to 32bit targets and add oC
26876 constraint for 64bit targets.
26877 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
26878 (*pushdf): Change rmF constraint to rmC.
26880 2017-03-03 Martin Liska <mliska@suse.cz>
26882 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
26883 Remove unused variable.
26885 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26888 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
26889 is a memory operand, increase num_memory.
26890 (ix86_expand_args_builtin): Likewise.
26892 2017-03-03 Jan Hubicka <jh@suse.cz>
26895 * ipa-devirt.c (maybe_record_node): Properly handle
26896 __cxa_pure_virtual visibility.
26898 2017-03-03 Martin Liska <mliska@suse.cz>
26900 PR tree-optimization/79803
26901 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
26903 (pass_loop_prefetch::execute): Disabled optimization if an
26904 assumption about L1 cache size is not met.
26906 2017-03-03 Martin Liska <mliska@suse.cz>
26908 PR rtl-optimization/79574
26909 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
26910 (hash_scan_set): Likewise.
26911 (dump_hash_table): Likewise.
26912 (hoist_code): Likewise.
26914 2017-03-03 Richard Biener <rguenther@suse.de>
26916 * fixed-value.c (fixed_from_string): Restore use of elt (1)
26917 in place of uhigh ().
26918 (fixed_convert_from_real): Likewise.
26920 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26923 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
26925 2017-03-03 Richard Biener <rguenther@suse.de>
26927 PR middle-end/79818
26928 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
26929 TYPE_OVERFLOW_UNDEFINED check.
26931 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26933 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
26935 (vector_ae_<mode>_p): Likewise.
26936 (vector_nez_<mode>_p): Likewise.
26937 (vector_ne_v2di_p): Likewise.
26938 (vector_ae_v2di_p): Likewise.
26939 (vector_ne_<mode>_p): Likewise.
26940 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
26942 (vsx_tsqrt<mode>2_fe): Likewise.
26944 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
26947 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
26949 2017-03-02 Jakub Jelinek <jakub@redhat.com>
26951 PR rtl-optimization/79780
26952 * cprop.c (one_cprop_pass): When second and further conditional trap
26953 in a single basic block is turned into an unconditional trap, turn it
26954 into a deleted note to avoid RTL verification failures.
26956 2017-03-02 Richard Biener <rguenther@suse.de>
26958 * fold-const.c (const_binop): Use ulow () instead of elt (0).
26960 2017-03-02 Richard Biener <rguenther@suse.de>
26962 PR tree-optimization/79345
26964 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
26965 param and abort the walk, returning -1 if it is hit.
26966 (walk_aliased_vdefs): Take a limit param and pass it on.
26967 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
26968 defaulting to 0 and return a signed int.
26969 * tree-ssa-uninit.c (struct check_defs_data): New struct.
26970 (check_defs): New helper.
26971 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
26972 about uninitialized memory.
26973 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
26974 bogus uninitialized warning.
26975 (fixed_convert_from_real): Likewise.
26977 2017-03-02 Bin Cheng <bin.cheng@arm.com>
26979 PR tree-optimization/66768
26980 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
26981 iv_use if base object can't be determined.
26983 2017-03-02 Jakub Jelinek <jakub@redhat.com>
26985 PR tree-optimization/79345
26986 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
26987 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
26988 (get_pattern_stats): Initialize it.
26989 * genemit.c (gen_expand): Verify match_scratch numbers come after
26990 match_operand/match_dup numbers.
26991 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
26992 match_scratch numbers.
26993 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
26995 * config/s390/s390.md (trunctdsd2): Likewise.
26997 2017-03-02 Richard Biener <rguenther@suse.de>
26999 * wide-int.h (wide_int_storage::operator=): Implement in terms
27002 2017-03-02 Richard Biener <rguenther@suse.de>
27004 PR tree-optimization/79777
27005 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
27006 the to insert expression to sth existing.
27008 2017-03-01 Martin Sebor <msebor@redhat.com>
27010 PR middle-end/79692
27011 * gimple-ssa-sprintf.c
27012 (directive::known_width_and_precision): New function.
27013 (format_integer): Use it.
27014 (get_mpfr_format_length): Consider the full range of precision
27015 when computing %g output with the # flag. Set the likely byte
27016 count to 3 rather than 1 when precision is indeterminate.
27017 (format_floating): Correct the lower bound of precision.
27019 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27021 * doc/invoke.texi: Document default code model for 64-bit Linux.
27023 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27026 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
27027 udiv rather than div since input pattern is unsigned.
27029 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
27031 * config/i386/i386.c (print_reg): Warn for values of
27032 unsupported size in integer register.
27034 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
27037 * config/rs6000/predicates.md (current_file_function_operand): Do
27038 not allow self calls to be local if the function is replaceable.
27040 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
27043 * config/rs6000/altivec.h (vec_ctz and others): Change the
27044 preprocessor macro that controls conditional compilation from
27045 _ARCH_PWR9 to __POWER9_VECTOR__.
27046 (vec_all_ne): Change parameterization of __altivec_scalar_pred
27047 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
27048 control (instead of _ARCH_PWR9 control) so that template
27049 definition uses power9-specific function.
27050 (vec_any_eq): Likewise.
27051 (vec_all_ne): Change macro definition to use a power9-specific
27052 expansion under #ifdef __POWER9_VECTOR__ control (instead of
27053 _ARCH_PWR9 control).
27054 (vec_any_eq) Likewise.
27055 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
27056 expansion for CMPNEF to remove support for xvcmpnesp instruction.
27057 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
27058 support for xvcmpnedp instruction.
27059 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
27060 macro expansion so that Power9 implementation of vec_all_ne does
27061 not use the AltiVec predicate framework.
27062 (VCMPNEH_P): Likewise.
27063 (VCMPNEW_P): Likewise.
27064 (VCMPNED_P): Likewise.
27065 (VCMPNEFP_P): Likewise.
27066 (VCMPNEDP_P): Likewise.
27067 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
27068 implementation of vec_any_eq to not use AltiVec predicate
27070 (VCMPAEH_P): Likewise.
27071 (VCMPAEW_P): Likewise.
27072 (VCMPAED_P): Likewise.
27073 (VCMPAEFP_P): Likewise.
27074 (VCMPAEDP_P): Likewise.
27075 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
27076 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
27077 not use the AltiVec predicate framework.
27078 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
27079 of vec_any_eq to not use AltiVec predicate framework.
27080 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27081 support for predefined __POWER9_VECTOR__ macro to indicate that
27082 Power9 instruction selection is enabled.
27083 (altivec_overloaded_builtins): Remove extraneous
27084 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
27085 function argument types RS6000_BTI_bool_V16QI and
27086 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
27087 entry for overloaded function argument types RS6000_BTI_bool_V4SI
27088 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
27089 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
27090 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
27091 Power9 for implementations of vec_cmpne. Change the signature for
27092 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
27093 (representing vec_all_ne) to remove the previously described first
27094 argument of type RS6000_BTI_INTSI, as this was an artifact of
27095 reliance on the AltiVec predicate framework, which is no longer
27096 used in the implementation of these functions. Add
27097 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
27098 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
27099 since, unlike the AltiVec predicate framework implementation, we
27100 do not share function descriptors between vec_alle and vec_anyeq.
27101 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
27102 set of modes that receive special treatment even when
27103 TARGET_P9_VECTOR is true. The special treatment emits code that
27104 does not depend on Power9 instructions.
27105 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
27106 define_expand to not rely on AltiVec predicate framework.
27107 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
27109 (vector_ne_v2di_p): Change this define_expand to not rely on
27110 AltiVec predicate framework.
27111 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
27113 (vector_ne_<mode>_p): Change this define_expand to not rely on
27114 AltiVec predicate framework.
27115 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
27117 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
27118 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
27119 define_insn pattern.
27120 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
27121 define_insn pattern because the xvcmpne<VSs>. instruction is not
27123 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
27124 instruction is not supported.
27126 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27128 * config/nvptx/nvptx.c: Include intl.h.
27130 2017-03-01 Martin Jambor <mjambor@suse.cz>
27133 * ipa-prop.h (ipa_bits): Removed field known.
27134 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
27135 to pointers. Adjusted their comments to warn about their sharing.
27136 (ipcp_transformation_summary): Change bits to a vector of pointers.
27137 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
27138 (ipa_get_ipa_bits_for_value): Declare.
27139 * tree-vrp.h (value_range): Mark as GTY((for_user)).
27140 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
27141 (ipa_bits_hash_table): Likewise.
27142 (ipa_vr_ggc_hash_traits): Likewise.
27143 (ipa_vr_hash_table): Likewise.
27144 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
27145 being pointers and vr_known being removed.
27146 (ipa_set_jf_unknown): Likewise.
27147 (ipa_get_ipa_bits_for_value): New function.
27148 (ipa_set_jfunc_bits): Likewise.
27149 (ipa_get_value_range): New overloaded functions.
27150 (ipa_set_jfunc_vr): Likewise.
27151 (ipa_compute_jump_functions_for_edge): Use the above functions to
27152 construct bits and vr parts of jump functions.
27153 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
27154 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
27156 (ipcp_grow_transformations_if_necessary): Also allocate
27157 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
27159 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
27160 them. Fix too long lines.
27161 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
27162 vr_known being removed.
27163 (ipa_read_jump_function): Use new setter functions to construct bits
27164 and vr parts of jump functions or set them to NULL.
27165 (write_ipcp_transformation_info): Adjust for bits being pointers.
27166 (read_ipcp_transformation_info): Likewise.
27167 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
27169 Include gt-ipa-prop.h.
27170 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
27172 (ipcp_store_bits_results): Likewise.
27173 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
27174 Do not write to existing jump functions but use a temporary instead.
27176 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27179 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
27180 attempt to use its first operand as BIT_FIELD_REF base.
27182 2017-03-01 Richard Biener <rguenther@suse.de>
27184 PR middle-end/79721
27185 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
27186 interpolating formula in wrapping arithmetic.
27187 (chrec_apply): Convert chrec_evaluate return value to wanted type.
27189 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27191 PR tree-optimization/79734
27192 * tree-vect-generic.c (expand_vector_condition): Optimize
27193 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
27194 Handle VEC_COND_EXPR where comparison has different inner width from
27195 type's inner width.
27197 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
27199 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
27200 markup, and similar issues. Remove @opindex entries for things
27201 that aren't options. Add missing -mmpy-option entries.
27203 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27205 PR tree-optimization/79737
27206 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
27207 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
27208 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
27209 instead of byte_size. Formatting fix.
27210 (shift_bytes_in_array_right): Formatting fix.
27212 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
27215 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
27216 condition on optimize for the leaf function test.
27218 2017-02-28 Martin Liska <mliska@suse.cz>
27221 * read-rtl-function.c (function_reader::handle_unknown_directive):
27222 Bail out when one uses -flto.
27224 2017-02-28 Martin Liska <mliska@suse.cz>
27226 * common.opt: Replace space with tabular for options of <number>
27228 * config/i386/i386.opt: Show <number> value for
27229 -mlarge-data-threshold.
27230 * opts.c (print_filtered_help): Do not display number in hexadecimal
27233 2017-02-28 Martin Liska <mliska@suse.cz>
27235 * common.opt: Fix --help=option -Q for options which are of
27238 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
27240 * config/i386/i386.c (print_reg): Error out for values
27241 of 8-bit size in invalid integer register.
27243 2017-02-28 Martin Sebor <msebor@redhat.com>
27245 PR tree-optimization/79691
27246 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
27248 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27251 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
27252 gcc_unreachable with output_operand_lossage.
27254 2017-02-28 Richard Biener <rguenther@suse.de>
27256 PR tree-optimization/79740
27257 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
27259 (visit_nary_op): Insert the nary into the hashtable if we
27260 pattern-matched sth.
27261 * tree-ssa-pre.c (eliminate_insert): Robustify.
27263 2017-02-28 Richard Biener <rguenther@suse.de>
27265 PR middle-end/79731
27266 * fold-const.c (decode_field_reference): Reject out-of-bound
27269 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27271 * config/i386/i386.c: Include intl.h.
27272 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
27273 instead of just cond ? "..." : "...".
27274 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
27275 * coverage.c (read_counts_file): Likewise.
27276 * omp-offload.c: Include intl.h.
27277 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
27278 of just cond ? "..." : "...".
27279 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
27280 of just cond ? "..." : "...".
27282 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
27285 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
27287 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
27288 'tune for' CPU name.
27289 * config/arm/arm-cpu-data.h: Regenerated.
27291 2017-02-28 Richard Biener <rguenther@suse.de>
27293 PR tree-optimization/79732
27294 * tree-inline.c (expand_call_inline): Do not shadow var.
27296 2017-02-28 Richard Biener <rguenther@suse.de>
27298 PR tree-optimization/79723
27299 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
27300 address-space properly.
27302 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
27304 * doc/optinfo.texi (Optimization groups): Fix option used for
27306 * doc/invoke.texi (-fopt-info): Document "omp".
27307 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
27308 (OPTGROUP_ALL): Add OPTGROUP_OMP.
27309 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
27310 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
27311 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
27313 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
27315 * dumpfile.c (optgroup_options): Instead of "openmp", associate
27316 OPTGROUP_OMP with "omp".
27318 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
27321 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
27322 for arithmetic shift of unsigned V2DI.
27324 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
27326 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
27327 arc/linux.h headers.
27328 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
27329 (LINK_SPEC): Likewise.
27330 (ARC_TLS_EXTRA_START_SPEC): Likewise.
27331 (EXTRA_SPECS): Likewise.
27332 (STARTFILE_SPEC): Likewise.
27333 (ENDFILE_SPEC): Likewise.
27334 (LIB_SPEC): Likewise.
27335 (TARGET_SDATA_DEFAULT): Likewise.
27336 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
27337 (MULTILIB_DEFAULTS): Likewise.
27338 (DWARF2_UNWIND_INFO): Likewise.
27339 * config/arc/big.h: New file.
27340 * config/arc/elf.h: Likewise.
27341 * config/arc/linux.h: Likewise.
27342 * config/arc/t-uClibc: Remove.
27344 2017-02-27 Bin Cheng <bin.cheng@arm.com>
27346 PR tree-optimization/77536
27347 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
27348 (tree_transform_and_unroll_loop): Use above function to compute the
27349 estimated niter of unrolled loop and use it when scaling profile.
27350 Also use count info rather than frequency if it's non-zero.
27351 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
27352 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
27353 (vect_transform_loop): Call above function.
27355 2017-02-27 Richard Biener <rguenther@suse.de>
27357 PR tree-optimization/45397
27358 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
27359 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
27360 (visit_nary_op): Add pattern matching for CSEing sign-changed
27361 or truncated operations with wider ones.
27363 2017-02-27 Richard Biener <rguenther@suse.de>
27365 PR tree-optimization/79690
27366 * tree-vect-stmts.c (vectorizable_store): Use vector type
27367 built from the DR with address-space.
27369 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
27371 * doc/invoke.texi (Optimize Options): Refine the description
27372 of asan-use-after-return.
27374 2017-02-25 Alan Modra <amodra@gmail.com>
27376 PR rtl-optimization/79584
27377 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
27378 base, not ad->base_term, the reg within base. Remove assertion
27379 that ad->base == ad->base_term. Replace gen_int_mode using
27380 bogus mode with const0_rtx.
27382 2017-02-25 Jakub Jelinek <jakub@redhat.com>
27384 PR middle-end/79396
27385 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
27386 FMA_EXPR like tcc_binary or tcc_unary.
27388 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
27391 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
27393 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
27394 (output_loc_operands): Handle DW_OP_call_ref and
27395 DW_OP_GNU_variable_value.
27396 (struct variable_value_struct): New type.
27397 (struct variable_value_hasher): Likewise.
27398 (variable_value_hash): New variable.
27399 (string_types): Remove.
27400 (copy_loc_descr): New function.
27401 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
27402 (prepend_loc_descr_to_each): New function.
27403 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
27404 instead of add_loc_descr_to_each if the first argument is single
27405 location list and the second has multiple.
27406 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
27407 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
27408 when looking for variable value which doesn't have other location info.
27409 (loc_list_from_tree): Formatting fix.
27410 (gen_array_type_die): Simplify DW_AT_string_length handling.
27411 (adjust_string_types): Remove.
27412 (gen_subprogram_die): Don't call adjust_string_types nor test/set
27413 string_types. Call resolve_variable_values.
27414 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
27415 (resolve_addr_in_expr): Likewise. Add A argument.
27416 (copy_deref_exprloc): Remove deref argument. Adjust for the
27417 original expression being DW_OP_GNU_variable_value with optionally
27418 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
27419 optionally after it.
27420 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
27421 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
27422 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
27423 (variable_value_hasher::hash, variable_value_hasher::equal): New
27425 (resolve_variable_value_in_expr, resolve_variable_value,
27426 resolve_variable_values, note_variable_value_in_expr,
27427 note_variable_value): New functions.
27428 (dwarf2out_early_finish): Call note_variable_value on all toplevel
27431 2017-02-24 Jakub Jelinek <jakub@redhat.com>
27434 * opts.h (handle_generated_option): Add GENERATED_P argument.
27435 * opts-common.c (handle_option): Adjust function comment.
27436 (handle_generated_option): Add GENERATED_P argument, pass it to
27438 (control_warning_option): Pass false to handle_generated_option
27440 * opts.c (maybe_default_option): Pass true to handle_generated_option
27442 * optc-gen.awk: Likewise.
27444 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27446 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
27447 a REG, look at the REG it is a SUBREG of.
27448 (splitter for cmpeqsi_t): Ditto.
27450 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27452 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
27453 the special USEs with the pattern of the insn, not the insn itself.
27455 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
27458 * doc/invoke.texi: Document -mload-store-pairs.
27460 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27461 Sandra Loosemore <sandra@codesourcery.com>
27463 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
27464 argument isn't a CONST_INT.
27465 (nios2_alternate_compare_const): Assert op is a CONST_INT.
27466 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
27467 (nios2_validate_compare): Bypass alternate compare logic if *op2
27468 is not a CONST_INT.
27469 (ldstwm_operation_p): Return false if first_base is not a REG or
27470 if first_offset is not a CONST_INT.
27472 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27474 * config/cris/cris.md: Use correct operand in a define_peephole2.
27476 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27478 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
27480 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27482 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
27483 this_insn if it is an INSN or JUMP_INSN.
27484 (force_offsettable): Look at base, not at addr.
27485 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
27486 on things that aren't necessarily CONST_INTs.
27488 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
27490 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
27491 -mfpmath=sse is the default also for x86-32 targets with SSE2
27492 instruction set when @option{-ffast-math} is enabled
27494 2017-02-24 Jeff Law <law@redhat.com>
27496 PR rtl-optimizatoin/79286
27497 * ira.c (update_equiv_regs): Drop may_trap_p exception to
27500 2017-02-24 Richard Biener <rguenther@suse.de>
27502 PR tree-optimization/79389
27503 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
27506 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
27508 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
27509 function comment to reflect reality.
27510 (loop_exits_before_overflow): Fix typo in function description.
27512 2017-02-24 Richard Biener <rguenther@suse.de>
27514 PR tree-optimization/79389
27515 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
27516 properly that a threading opportunity exists. Detect conditional
27517 copy/constant propagation opportunities.
27519 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
27521 * config/visium/visium.md (type): Add trap.
27522 (b): New mode attribute.
27523 (*btst): Rename into...
27524 (*btst<mode>): ...this and adjust.
27525 (*cbranchsi4_btst_insn): Rename into...
27526 (*cbranch<mode>4_btst_insn): ...this and adjust.
27527 (trap): New define_insn.
27529 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27531 PR tree-optimization/79389
27532 * ifcvt.c (struct noce_if_info): Add rev_cond field.
27533 (noce_reversed_cond_code): New function.
27534 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
27535 reversed_comparison_code. Formatting fix.
27536 (noce_try_store_flag): Test rev_cond != NULL in addition to
27537 reversed_comparison_code.
27538 (noce_try_store_flag_constants): Likewise.
27539 (noce_try_store_flag_mask): Likewise.
27540 (noce_try_addcc): Use rev_cond if non-NULL instead of
27541 reversed_comparison_code.
27542 (noce_try_cmove_arith): Likewise. Formatting fixes.
27543 (noce_try_minmax, noce_try_abs): Clear rev_cond.
27544 (noce_find_if_block): Initialize rev_cond.
27545 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
27546 instead of false as last argument never attempt to reverse it
27549 2017-02-23 Bin Cheng <bin.cheng@arm.com>
27551 PR tree-optimization/79663
27552 * tree-predcom.c (combine_chains): Process refs in reverse order
27553 only for ZERO length chains, and add explaining comment.
27555 2017-02-23 Jeff Law <law@redhat.com>
27557 PR tree-optimization/79578
27558 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
27559 in call to operand_equal_p.
27561 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
27564 * config/i386/cpuid.h: Fix another undefined behavior.
27566 2017-02-23 Richard Biener <rguenther@suse.de>
27568 PR tree-optimization/79683
27569 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
27570 vector types for data-refs.
27572 2017-02-23 Martin Liska <mliska@suse.cz>
27574 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
27576 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27578 PR middle-end/79665
27579 * internal-fn.c (get_range_pos_neg): Moved to ...
27580 * tree.c (get_range_pos_neg): ... here. No longer static.
27581 * tree.h (get_range_pos_neg): New prototype.
27582 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
27583 are known to be in between 0 and signed maximum inclusive, try to
27584 expand both unsigned and signed divmod and use the cheaper one from
27587 2017-02-22 Jeff Law <law@redhat.com>
27589 PR tree-optimization/79578
27590 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
27591 to compare base operands.
27593 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
27596 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
27597 gpc_reg_operand instead of fpr_reg_operand.
27599 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
27601 * config/mips/mips.c (mips_return_in_memory): Force FP
27602 vector types to be returned in memory for o32 ABI.
27604 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27606 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
27607 instead of DW_TAG_member for static data member declarations and don't
27608 set no_linkage_name for static inline data members.
27609 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
27612 2017-02-22 Martin Liska <mliska@suse.cz>
27614 * doc/invoke.texi: Replace inequality signs with square brackets
27617 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27619 PR tree-optimization/68644
27620 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
27622 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
27625 * lra-constraints.c (simplify_operand_subreg): Handle
27626 WORD_REGISTER_OPERATIONS targets.
27628 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27631 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
27632 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
27633 elimination by swapping fld*.
27635 2017-02-22 Richard Biener <rguenther@suse.de>
27637 PR tree-optimization/79673
27638 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
27639 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
27640 irrelevant address-space qualifiers and avoiding a
27641 ADDR_SPACE_CONVERT_EXPR from fold_convert.
27643 2017-02-22 Richard Biener <rguenther@suse.de>
27645 PR tree-optimization/79666
27646 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
27647 to not symbolically negate if that may introduce undefined
27650 2017-02-22 Martin Liska <mliska@suse.cz>
27653 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
27654 * data-streamer-out.c (streamer_write_gcov_count_stream):
27656 * value-prof.c (stream_out_histogram_value): Make assert more
27657 precise based on type of counter.
27659 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
27662 * config/i386/i386.md (standard_x87sse_constant_load splitter):
27663 Use nonimmediate_operand instead of memory_operand for operand 1.
27664 (float-extend standard_x87sse_constant_load splitter): Ditto.
27666 2017-02-21 Jeff Law <law@redhat.com>
27668 PR tree-optimization/79621
27669 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
27670 blocks with edges to themselves.
27672 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27675 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
27676 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
27677 Use gimple_call_builtin_p.
27680 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
27681 on temporarily removed DEBUG_INSNs.
27683 PR tree-optimization/79649
27684 * tree-loop-distribution.c (classify_partition): Give up on
27685 non-generic address space loads/stores.
27687 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
27689 * doc/loop.texi (Loop manipulation): Remove nonexistent
27690 tree_ssa_loop_version from the documentation.
27691 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
27693 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27696 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
27697 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
27698 * config/rs6000/rs6000.c: Include except.h.
27699 (rs6000_expand_split_stack_prologue): Call
27700 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
27702 2017-02-21 Martin Jambor <mjambor@suse.cz>
27705 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
27706 have been analyzed.
27708 2017-02-21 Martin Jambor <mjambor@suse.cz>
27710 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
27711 for backward compatibility only.
27712 * doc/invoke.texi (Option Summary): Remove all references to
27713 -fipa-cp-alignment.
27715 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
27719 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27721 * lra-constraints.c (curr_insn_transform): Handle
27722 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27724 2017-02-21 Martin Liska <mliska@suse.cz>
27726 * config/i386/i386.opt: Replace -masm-dialect with -masm.
27728 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
27730 PR translation/79638
27731 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
27733 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
27736 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
27737 (arm_function_ok_for_sibcall): Return false for an indirect call by
27738 descriptor if all the argument registers are used.
27739 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
27740 alignment of the function.
27742 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27744 PR tree-optimization/61441
27745 * simplify-rtx.c (simplify_const_unary_operation): For
27746 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
27747 the sNaN unmodified.
27749 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27751 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27752 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27753 instead of SYSTEM_HEADER_DIR.
27755 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
27756 Martin Liška <mliska@suse.cz>
27758 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
27759 Fix typos and grammar, use active voice, and clarify.
27761 2017-02-20 Marek Polacek <polacek@redhat.com>
27763 PR middle-end/79537
27764 * gimplify.c (gimplify_expr): Handle unused *&&L;.
27767 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
27769 2017-02-20 Jakub Jelinek <jakub@redhat.com>
27772 * config/i386/i386.c (ix86_expand_builtin): Handle
27773 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
27774 ix86_builtins_isa[fcode].isa as a requirement of those
27775 flags and any other flag in the bitmask.
27776 (ix86_init_mmx_sse_builtins): Use 0 instead of
27777 ~OPTION_MASK_ISA_64BIT as mask.
27778 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
27779 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
27780 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
27781 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
27783 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27786 * lra-constraints.c (split_reg): Check requested split mode
27787 is supported by the register.
27789 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27791 * lra-constraints.c (simplify_operand_subreg): Remove early
27794 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27797 * lra-constraints.c (curr_insn_transform): Tighten condition
27798 for converting SUBREG reloads from OP_OUT to OP_INOUT.
27800 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27803 * lra-constraints.c (curr_insn_transform): Handle
27804 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27806 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
27809 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
27811 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
27813 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
27816 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
27819 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27821 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
27824 2017-02-18 Jakub Jelinek <jakub@redhat.com>
27826 * final.c (last_columnnum, override_columnnum): New variables.
27827 (final_start_function): Set last_columnnum, pass it to begin_prologue
27828 hook and pass 0 to dwarf2out_begin_prologue.
27829 (final_scan_insn): Update override_columnnum. Pass last_columnnum
27830 to source_line debug hook.
27831 (notice_source_line): Compute last_columnnum and for debug_column_info
27832 return true on column changes.
27833 * debug.h (struct gcc_debug_hooks): Add column argument to
27834 source_line and begin_prologue hooks.
27835 (debug_nothing_int_charstar_int_bool): Remove prototype.
27836 (debug_nothing_int_int_charstar,
27837 debug_nothing_int_int_charstar_int_bool): New prototypes.
27838 (dwarf2out_begin_prologue): Add column argument.
27839 * debug.c (do_nothing_debug_hooks): Adjust source_line and
27840 begin_prologue hooks.
27841 (debug_nothing_int_charstar_int_bool): Remove.
27842 (debug_nothing_int_int_charstar,
27843 debug_nothing_int_int_charstar_int_bool): New functions.
27844 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
27845 through to dwarf2out_source_line.
27846 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
27847 (dwarf2out_source_line): Add column argument, emit it if requested.
27848 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
27850 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27851 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27852 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
27853 through to dwarf2out_begin_prologue.
27854 (vmsdbgout_source_line): Add column argument, pass it through to
27855 dwarf2out_source_line.
27856 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
27857 dbxout_source_line caller.
27858 (dbxout_source_line): Add column argument.
27860 * common.opt (gno-column-info, gcolumn-info): New options.
27861 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
27862 (check_die): Also test for multiple DW_AT_decl_column attributes.
27863 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
27864 DW_AT_decl_column if requested.
27865 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
27867 (gen_variable_die): Likewise.
27868 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
27869 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
27872 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
27873 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
27874 (ix86_handle_option): Handle OPT_m3dnowa.
27875 * doc/invoke.texi (-m3dnowa): Document.
27876 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
27877 -m3dnowa instead of -m3dnow -march=athlon.
27880 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
27881 instead of gcc_assert for K, r and R code checks. Formatting fixes.
27883 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27886 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
27887 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
27888 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
27889 generator for vsx_xxpermdi_<mode>_be.
27890 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
27891 force big-endian semantics.
27892 (vsx_xxpermdi_<mode>_be): New define_expand with same
27893 implementation as previous version of vsx_xxpermdi_<mode>.
27895 2017-02-17 Jakub Jelinek <jakub@redhat.com>
27897 PR tree-optimization/79327
27898 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
27899 variable, its initialization and use.
27901 2017-02-17 Julia Koval <julia.koval@intel.com>
27903 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
27904 (OPTION_MASK_ISA_PKU_UNSET): New.
27905 (ix86_handle_option): Handle -mrdpid.
27906 * config/i386/cpuid.h (bit_RDPID): New.
27907 * config/i386/driver-i386.c (host_detect_local_cpu):
27908 Detect RDPID feature.
27909 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
27910 * config/i386/i386-c.c (ix86_target_macros_internal):
27912 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
27913 (ix86_valid_target_attribute_inner_p): Add "rdpid".
27914 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
27915 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
27916 * config/i386/i386.md (define_insn "rdpid"): New.
27917 * config/i386/i386.opt Add -mrdpid.
27918 * config/i386/immintrin.h (_rdpid_u32): New.
27920 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
27922 PR rtl-optimization/79541
27923 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
27924 instead of transforming it into USE.
27926 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
27928 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
27929 If HONOR_SNANS (SFmode) force the input to a register.
27930 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
27931 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
27932 an frsp or similar insn.
27934 2017-02-17 Martin Liska <mliska@suse.cz>
27936 PR rtl-optimization/79577
27937 * params.def (selsched-max-sched-times): Increase minimum to 1.
27939 2017-02-17 Martin Liska <mliska@suse.cz>
27941 PR rtl-optimization/79574
27942 * gcse.c (want_to_gcse_p): Prevent integer overflow.
27944 2017-02-17 Martin Liska <mliska@suse.cz>
27946 PR tree-optimization/79529
27947 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
27948 ssa_defined_default_def_p to handle cases which are implicitly
27950 * tree-ssa.c (ssa_defined_default_def_p): New function.
27951 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
27952 which are implicitly defined.
27953 * tree-ssa.h (ssa_defined_default_def_p): Declare.
27955 2017-02-17 Richard Biener <rguenther@suse.de>
27957 PR middle-end/79576
27958 * params.def (max-ssa-name-query-depth): Limit to 10.
27960 2017-02-17 Richard Biener <rguenther@suse.de>
27962 PR tree-optimization/79552
27963 * tree-ssa-structalias.c (visit_loadstore): Properly verify
27966 2017-02-17 Richard Biener <rguenther@suse.de>
27969 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
27971 2017-02-17 Marek Polacek <polacek@redhat.com>
27973 PR middle-end/79536
27974 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
27975 (fold_negate_expr): New wrapper.
27977 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
27979 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
27980 Correct terminology and de-emphasize pre-standard behavior.
27982 2017-02-16 Alan Modra <amodra@gmail.com>
27984 PR rtl-optimization/79286
27985 * ira.c (def_dominates_uses): New function.
27986 (update_equiv_regs): Don't create an equivalence for insns that
27987 may trap where the register def does not dominate the use.
27989 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
27991 PR rtl-optimization/78127
27992 * lra.c (lra): Call lra_eliminate before finish the loop after
27995 2017-02-16 Richard Biener <rguenther@suse.de>
27997 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
27999 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
28000 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
28001 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
28002 (isl_val_int_from_wi): New function.
28003 (extract_affine_gmp): Rename to ...
28004 (extract_affine_wi): ... this, take a widest_int.
28005 (extract_affine_int): Just wrap extract_affine_wi.
28006 (add_param_constraints): Use isl_val_int_from_wi.
28007 (add_loop_constraints): Likewise, and extract_affine_wi.
28009 2017-02-15 Jeff Law <law@redhat.com>
28011 PR middle-end/79521
28012 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
28013 ira_init_register_move_cost_if_necessary.
28015 2017-02-15 Martin Sebor <msebor@redhat.com>
28017 PR middle-end/32003
28018 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
28019 removed in a prior commit.
28021 2017-02-15 Bin Cheng <bin.cheng@arm.com>
28023 PR tree-optimization/79347
28024 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
28025 counters during peeling.
28027 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
28029 * Makefile.in (site.exp): Remove "set ISLVER".
28031 2017-02-15 Jakub Jelinek <jakub@redhat.com>
28034 * real.c (real_from_integer): Call real_convert even for decimal.
28036 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
28039 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
28041 2017-02-14 Andrew Pinski <apinski@cavium.com>
28043 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
28044 cores and change the partno/implementer to be correct.
28045 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
28046 the 'B" as the implementer.
28047 * config/aarch64/aarch64-tune.md: Regenerate.
28049 2017-02-14 Carl Love <cel@us.ibm.com>
28051 * config/rs6000/rs6000.c: Add case statement entry to make the
28052 xvcvuxdsp built-in argument unsigned.
28053 * config/rs6000/vsx.md: Fix the source and return operand types so they
28054 match the instruction definitions from the ISA document. Fix typo
28055 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
28058 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
28061 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
28062 member early_clobber_alts.
28063 * lra-lives.c (reg_early_clobber_p): New.
28064 (process_bb_lives): Use it.
28065 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
28066 (debug_operand_data): Initialize early_clobber_alts.
28067 (setup_operand_alternative): Set up early_clobber_alts.
28068 (collect_non_operand_hard_regs): Ditto. Pass early clobber
28069 alternatives to new_insn_reg.
28070 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
28072 (lra_update_insn_regno_info): Pass the new arg.
28074 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28076 PR middle-end/79505
28077 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
28078 (new_oacc_loop_raw): Don't clear already cleared fields.
28081 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
28082 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
28083 _mm512_prefetch_i64gather_ps): New inline functions and macros.
28085 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
28088 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
28090 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
28093 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
28094 the extra instruction to the right place to store 128-bit constant
28097 2017-02-14 Martin Sebor <msebor@redhat.com>
28099 PR middle-end/79448
28100 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
28101 warning for strings of unknown length.
28103 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
28105 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
28107 2017-02-14 Jeff Law <law@redhat.com>
28110 * ira-costs.c (scan_one_insn): Initialize register move costs
28111 for pseudos seen in USE/CLOBBER insns.
28113 PR tree-optimization/79095
28114 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
28115 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
28116 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
28117 if the operands are known to be not equal, then the resulting range
28119 (intersect_ranges): If the new range is ~[0,0] and the old range is
28120 wide, then prefer ~[0,0].
28121 * tree-vrp.c (overflow_comparison_p_1): New function.
28122 (overflow_comparison_p): New function.
28123 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
28124 if NAME is used in an overflow test.
28125 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
28126 overflow check that can be expressed as an equality test, then adjust
28127 ops to be that equality test.
28129 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28131 * config/s390/s390-builtin-types.def: Remove flags argument.
28132 * config/s390/s390.c (s390_init_builtins): Likewise.
28134 2017-02-14 Martin Liska <mliska@suse.cz>
28136 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
28137 vector. Fix trailing white spaces.
28139 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
28141 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
28144 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28146 PR rtl-optimization/68664
28147 * config/arm/arm.c (arm_sched_can_speculate_insn):
28148 New function. Declare prototype.
28149 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
28151 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28153 PR rtl-optimization/68664
28154 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
28156 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
28158 2017-02-14 Amit Pawar <amit.pawar@amd.com>
28160 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
28161 max skip bytes for function, loop and jump.
28163 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28165 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
28166 ABS_EXPR for gimple dump.
28168 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28171 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
28173 PR tree-optimization/79408
28174 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
28175 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
28176 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
28177 also if rhs1 is INTEGER_CST.
28179 2017-02-14 Richard Biener <rguenther@suse.de>
28181 PR middle-end/79432
28182 * tree-into-ssa.c (insert_phi_nodes): When the function can
28183 have abnormal edges rewrite SSA names with broken use-def
28184 dominance out of SSA and register them for PHI insertion.
28186 2017-02-13 Martin Sebor <msebor@redhat.com>
28188 PR middle-end/79496
28189 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
28190 clearing info.nowrite flag when snprintf size argument is a range.
28192 2017-02-13 Jakub Jelinek <jakub@redhat.com>
28194 * cprop.c (cprop_jump): Add missing space in string literal.
28195 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
28196 (get_constraint_for_component_ref): Likewise.
28197 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
28198 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
28199 * lra-constraints.c (process_alt_operands): Likewise.
28200 * ipa-inline.c (inline_small_functions): Likewise.
28201 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
28202 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28203 * trans-mem.c (diagnose_tm_1_op): Likewise.
28204 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
28205 (grid_parallel_clauses_gridifiable): Likewise.
28207 * config/nvptx/mkoffload.c (process): Add space in between
28210 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
28211 "MOD4_SSE_REGS" and "ALL_REGS".
28213 * spellcheck.c (test_data): Add , in between "foo" and "food".
28215 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28218 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
28219 boundary crossing check and subsequent code generation agree.
28221 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28223 * config/aarch64/aarch64.c (has_memory_op): Delete.
28224 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
28227 2017-02-13 Jakub Jelinek <jakub@redhat.com>
28229 PR rtl-optimization/79388
28230 PR rtl-optimization/79450
28231 * combine.c (distribute_notes): When removing TEM_INSN for which
28232 corresponding dest has last value recorded, invalidate that last
28235 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28237 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
28238 of explicit '@'. Add missing assembly comment marker on branch costs
28241 2017-02-13 Nathan Sidwell <nathan@acm.org>
28243 * gengtype-lex.l (<in_struct>): Add '/'.
28245 2017-02-13 Martin Liska <mliska@suse.cz>
28248 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
28250 2017-02-13 Richard Biener <rguenther@suse.de>
28252 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
28254 * configure: Re-generate.
28255 * config.in: Likewise.
28256 * graphite-dependences.c: Simplify as if
28257 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
28258 * graphite-isl-ast-to-gimple.c: Likewise.
28259 * graphite-optimize-isl.c: Likewise.
28260 * graphite-poly.c: Likewise.
28261 * graphite-sese-to-poly.c: Likewise.
28262 * graphite.h: Likewise.
28263 * toplev.c: Include isl/version.h and use isl_version () for
28264 printing the ISL version.
28265 * doc/install.texi: Update ISL requirement.
28267 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
28269 * doc/standards.texi (Standards): Update reference to
28272 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
28274 * doc/extend.texi (Named Address Spaces): sourceware.org now
28276 * doc/install.texi (Binaries): Ditto.
28279 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
28281 * doc/cpp.texi: Replace "stringify"/"stringification" with C
28282 standard terminology "stringize"/"stringizing" throughout.
28283 * doc/cppinternals.texi: Likewise.
28285 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
28287 * doc/extend.texi: Fix some spelling mistakes and typos.
28288 * doc/invoke.texi: Likewise.
28290 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
28293 * params.def (inline-min-speedup) Change from 10 to 8.
28295 2017-02-11 Jakub Jelinek <jakub@redhat.com>
28297 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
28300 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
28303 * ipa-inline-analysis.c (get_minimal_bb): New function.
28304 (record_modified): Use it.
28305 (remap_edge_change_prob): Handle also ancestor functions.
28307 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
28309 * doc/contrib.texi (Contributors): Remove broken link into
28310 the Mauve CVS repository.
28312 2017-02-11 Jakub Jelinek <jakub@redhat.com>
28314 PR middle-end/79454
28315 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
28316 result computation whenever lhs doesn't have vector mode, not
28317 just when it has BLKmode.
28319 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
28321 * doc/makefile.texi (profiledbootstrap): Refer to the
28322 installation instructions only in textual form.
28324 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28327 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
28329 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
28331 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
28332 (Specific): Update mingw-w64 reference.
28334 (Specific): Remove broken link to Renesas RX processor.
28336 2017-02-10 Richard Biener <rguenther@suse.de>
28338 * toplev.c (process_options): Do not mention obsolete graphite
28339 options when printing sorry message about missing graphite support.
28340 Mention -floop-nest-optimize.
28342 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
28344 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
28345 (vtst_p16): Likewise.
28346 (vtstq_p8): Likewise.
28347 (vtstq_p16): Likewise.
28349 (vtstq_p64): Likewise.
28350 * config/arm/arm_neon.h (vgetq_lane_p64): New.
28351 (vset_lane_p64): New.
28352 (vsetq_lane_p64): New.
28354 2017-02-10 Jakub Jelinek <jakub@redhat.com>
28356 PR tree-optimization/79411
28357 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
28358 stmt operands are SSA_NAMEs used in abnormal phis.
28359 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
28362 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
28365 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
28368 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
28370 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
28372 2017-02-09 Jakub Jelinek <jakub@redhat.com>
28374 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
28378 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
28379 not arbitrary TREE_CONSTANT.
28382 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
28383 "omp declare target link" attribute unless is_global_var.
28384 * omp-offload.c (find_link_var_op): Likewise.
28386 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
28387 Chung-Lin Tang <cltang@codesourcery.com>
28389 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
28391 (gimplify_adjust_omp_clauses): Don't delete TILE.
28392 (gimplify_omp_for): Deal with TILE.
28393 * internal-fn.c (expand_GOACC_TILE): New function.
28394 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
28396 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
28397 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
28399 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
28400 avoid DIV for outermost collapse var.
28401 (expand_oacc_for): Insert tile element loop as needed. Adjust.
28402 Remove out of date comments, fix whitespace.
28403 * omp-general.c (omp_extract_for_data): Deal with tiling.
28404 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
28405 adjust OLF_DIM_BASE value.
28406 (struct omp_for_data): Add tiling field.
28407 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
28408 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
28409 for auto loops. Remove default auto determining, moved to
28410 oacc_loop_fixed_partitions.
28411 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
28412 stmts, add e_mask field.
28413 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
28414 (oacc_thread_numbers): Use oacc_dim_call.
28415 (oacc_xform_tile): New.
28416 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
28417 (finish_oacc_loop): Adjust for ifns vector.
28418 (oacc_loop_discover_walk): Append loop abstraction sites to list,
28419 add case for GOACC_TILE fns.
28420 (oacc_loop_xform_loop): Delete.
28421 (oacc_loop_process): Iterate over call list directly, and add
28422 handling for GOACC_TILE fns.
28423 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
28425 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
28426 vector partitioning to outer loops. Assign 2 partitions to loops
28427 when available. Add TILE handling.
28428 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
28429 (execite_oacc_device_lower): Process GOACC_TILE fns,
28430 ignore unknown specs.
28431 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
28432 * tree.c (omp_clause_num_ops): Adjust TILE ops.
28433 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
28435 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
28437 * configure.ac (ACX_BUGURL): Update.
28438 * configure: Regenerate.
28440 2017-02-09 Richard Biener <rguenther@suse.de>
28442 PR tree-optimization/69823
28443 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
28444 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
28446 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28448 * config/arc/arc-c.def: Add __NPS400__ definition.
28449 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
28450 (TARGET_NPS400): Define.
28452 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28454 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
28456 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
28457 pointer, arch_info.
28458 (arc_cpu_types): Fill the arch_info field with a pointer into the
28459 arc_arch_types table.
28460 (arc_selected_cpu): Declare.
28461 * config/arc/arc.c (arc_selected_cpu): Make global.
28462 (arc_selected_arch): Delete.
28463 (arc_base_cpu): Delete.
28464 (arc_override_options): Remove references to deleted variables,
28465 update access to arch information.
28466 (ARC_OPT): Update access to arch information.
28467 (ARC_OPTX): Likewise.
28468 * config/arc/arc.h (arc_base_cpu): Remove declaration.
28469 (TARGET_ARC600): Update access to arch information.
28470 (TARGET_ARC601): Likewise.
28471 (TARGET_ARC700): Likewise.
28472 (TARGET_EM): Likewise.
28473 (TARGET_HS): Likewise.
28474 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
28477 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
28480 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
28481 condition/operands for integer GE/LE/GEU/LEU operations.
28483 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
28485 PR translation/79397
28486 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
28489 2017-02-08 Martin Jambor <mjambor@suse.cz>
28492 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
28493 whether allocation happened.
28494 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
28495 nothing was allocated.
28497 2017-02-08 Jakub Jelinek <jakub@redhat.com>
28499 PR tree-optimization/79408
28500 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
28501 constant, but SSA_NAME with a known integer range, use the minimum
28502 of that range instead of op1 to determine if modulo can be replaced
28503 with its first operand.
28505 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28507 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
28509 2017-02-08 Richard Biener <rguenther@suse.de>
28511 PR tree-optimization/71824
28512 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28513 Check all loops contained in the merged region.
28515 2017-02-07 Andrew Pinski <apinski@cavium.com>
28517 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
28519 2017-02-07 Andrew Pinski <apinski@cavium.com>
28521 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
28522 (thunderxt88): Likewise.
28523 (thunderxt81): Disable LSE and change v8.1 to v8.
28524 (thunderxt83): Likewise.
28526 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28527 Richard Biener <rguenther@suse.de>
28529 PR middle-end/79399
28530 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
28531 type from int to size_t.
28532 * ira-costs.c (struct_costs_size): Change type from int to size_t.
28534 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28536 PR rtl-optimization/79386
28537 * cprop.c (bypass_conditional_jumps): Initialize
28538 bypass_last_basic_block already before splitting bbs after
28539 unconditional traps...
28540 (bypass_conditional_jumps): ... rather than here.
28543 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
28544 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
28545 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
28546 fix -masm=intel patterns.
28548 2017-02-07 Richard Biener <rguenther@suse.de>
28550 PR tree-optimization/79256
28551 PR middle-end/79278
28552 * builtins.c (get_object_alignment_2): Use min_align_of_type
28553 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
28554 and ADJUST_FIELD_ALIGN.
28556 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
28558 * doc/tm.texi: Regenerate.
28559 * stor-layout.c (layout_decl): Adjust.
28560 (update_alignment_for_field): Likewise.
28561 (place_field): Likewise.
28562 (min_align_of_type): Likewise.
28563 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
28564 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
28565 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
28566 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
28567 * config/frv/frv.c (frv_adjust_field_align): Likewise.
28568 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
28569 * config/i386/i386.c (x86_field_alignment): Likewise.
28570 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
28571 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
28572 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
28573 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
28574 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
28575 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
28579 2017-01-30 Richard Biener <rguenther@suse.de>
28581 PR tree-optimization/79256
28582 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28583 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28586 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
28588 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
28589 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
28590 builtins to SImode and emit a zero-extend, if necessary.
28592 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28594 * docs/invoke.texi (RISC-V Options): Alphabetize.
28596 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28598 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
28601 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28603 * config/riscv/riscv.c: New file.
28604 * common/config/riscv/riscv-common.c: Likewise.
28605 * config.gcc: Likewise.
28606 * config/riscv/constraints.md: Likewise.
28607 * config/riscv/elf.h: Likewise.
28608 * config/riscv/generic.md: Likewise.
28609 * config/riscv/linux.h: Likewise.
28610 * config/riscv/multilib-generator: Likewise.
28611 * config/riscv/peephole.md: Likewise.
28612 * config/riscv/pic.md: Likewise.
28613 * config/riscv/predicates.md: Likewise.
28614 * config/riscv/riscv-builtins.c: Likewise.
28615 * config/riscv/riscv-c.c: Likewise.
28616 * config/riscv/riscv-ftypes.def: Likewise.
28617 * config/riscv/riscv-modes.def: Likewise.
28618 * config/riscv/riscv-opts.h: Likewise.
28619 * config/riscv/riscv-protos.h: Likewise.
28620 * config/riscv/riscv.h: Likewise.
28621 * config/riscv/riscv.md: Likewise.
28622 * config/riscv/riscv.opt: Likewise.
28623 * config/riscv/sync.md: Likewise.
28624 * config/riscv/t-elf-multilib: Likewise.
28625 * config/riscv/t-linux: Likewise.
28626 * config/riscv/t-linux-multilib: Likewise.
28627 * config/riscv/t-riscv: Likewise.
28628 * configure.ac: Likewise.
28629 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
28630 Waterman as RISC-V maintainers.
28631 * doc/install.texi: Add RISC-V entries.
28632 * doc/invoke.texi: Add RISC-V options section.
28633 * doc/md.texi: Add RISC-V constraints section.
28634 * configure: Regenerated.
28636 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
28639 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
28640 false values to be constant vectors with all 0 or all 1 bits set.
28641 (vcondu<mode><mode>): Likewise.
28642 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
28644 (fpmask_comparison_operator): Update comment.
28645 (vecint_comparison_operator): New predicate.
28646 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
28647 vector conditionals when the true and false values are constant
28648 vectors with all 0 bits or all 1 bits set.
28650 2017-02-06 Martin Sebor <msebor@redhat.com>
28652 PR tree-optimization/79376
28653 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
28655 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
28657 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
28658 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
28659 to simplify split condition.
28661 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28663 * omp-expand.c (oxpand_omp_atomic_fetch_op,
28664 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
28667 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
28669 PR rtl-optimization/68664
28670 * target.def (can_speculate_insn): New hook.
28671 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
28672 * doc/tm.texi: Regenerate.
28673 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
28674 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
28675 (rs6000_sched_can_speculate_insn): New function.
28677 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28679 PR tree-optimization/79284
28680 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
28681 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
28682 vectorizable_mask_load_store, vectorizable_operation,
28683 vect_is_simple_cond, get_same_sized_vectype): Use it instead
28684 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
28685 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
28686 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
28687 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
28688 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28689 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
28690 is_gimple_assign (stmt). Replace another such test with
28691 is_gimple_assign (stmt).
28693 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28696 * config/avr/avr.c (rtl-iter.h): Include it.
28697 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
28698 (avr_legitimate_combined_insn): ...and implementation.
28700 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
28702 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
28703 * config/s390/s390.c (s390_const_operand_ok)
28704 (s390_canonicalize_comparison, s390_extract_part)
28705 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
28706 (s390_contiguous_bitmask_p, s390_rtx_costs)
28707 (legitimize_pic_address): Likewise.
28708 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
28709 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
28710 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
28711 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
28712 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
28714 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28716 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
28717 REGNO($0) == REGNO($1).
28719 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28721 * config/s390/linux.h(SIZE_TYPE): Add comment.
28723 2017-02-06 Julian Brown <julian@codesourcery.com>
28724 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28725 Virendra Pathak <virendra.pathak@broadcom.com>
28727 * config/aarch64/aarch64-cores.def: Change the scheduler
28729 * config/aarch64/aarch64.md: Include thunderx2t99.md.
28730 * config/aarch64/thunderx2t99.md: New file.
28732 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28734 * doc/standards.texi (Go Language): Update link to language
28737 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28739 * tree-eh.c (lower_resx): Sanitize profile.
28740 (cleanup_empty_eh_move_lp): Likewise.
28742 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28745 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
28747 * cfgloopmanip.h (loop_version): Update prototype.
28748 * modulo-sched.c (sms_schedule): Update call of loop_version.
28749 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
28750 * tree-parloops.c (gen_parallel_loop): Likewise.
28751 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
28752 * tree-ssa-loop-split.c (split_loop): Likewise.
28753 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
28754 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
28756 2017-02-05 Martin Liska <mliska@suse.cz>
28759 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
28761 (print_operand_address): Initialize a struct to zero.
28763 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28765 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
28766 garbage collector only in textual form.
28768 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28770 * doc/extend.texi (x86 specific memory model extensions for
28771 transactional memory): Simplify a phrase.
28773 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
28776 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
28777 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
28778 (atomic_storedi_1): Likewise.
28780 2017-02-04 Jakub Jelinek <jakub@redhat.com>
28782 PR tree-optimization/79338
28783 * tree-parloops.c (gather_scalar_reductions): Don't call
28784 vect_analyze_loop_form for loop->inner before destroying loop's
28787 2017-02-03 Martin Sebor <msebor@redhat.com>
28789 PR tree-optimization/79327
28790 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
28791 when precision has resulted in leading zeros.
28792 (format_integer): Adjust the likely counter to assume an unknown
28793 argument that may be zero is non-zero.
28795 2017-02-03 Jason Merrill <jason@redhat.com>
28798 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
28799 avoid copying non-taken branch.
28801 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28803 PR tree-optimization/79340
28804 * tree-vect-loop.c (vectorizable_reduction): Release
28805 vec_defs elements after safe_splicing them into other vectors.
28808 PR tree-optimization/79327
28809 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
28810 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
28812 (format_integer): Use wide_int_to_tree instead of build_int_cst
28813 + to_?hwi. If argmin is NULL, just set argmin and argmax to
28814 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
28815 of shortest and longest sequence.
28817 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
28819 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
28820 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
28822 2017-02-03 Walter Lee <walt@tilera.com>
28825 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
28826 after initial stackframe link reg save.
28827 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
28829 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28832 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
28833 wu for stxssp alternative.
28835 2017-02-03 Martin Sebor <msebor@redhat.com>
28837 PR tree-optimization/79352
28838 * gimple-fold.c (get_range_strlen): Add argument.
28839 (get_range_strlen): Change return type to bool.
28840 (get_maxval_strlen): Pass in a dummy argument.
28841 * gimple-fold.h (get_range_strlen): Change return type to bool.
28842 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
28843 * tree.h (array_at_struct_end_p): Add argument.
28844 * tree.c (array_at_struct_end_p): Handle it.
28846 2017-02-03 Martin Liska <mliska@suse.cz>
28849 * multiple_target.c (create_dispatcher_calls): Redirect edge
28850 from a caller of a dispatcher.
28851 (expand_target_clones): Make the clones local.
28852 (ipa_target_clone): Do both target clones and resolvers.
28853 (ipa_dispatcher_calls): Remove the pass.
28854 (pass_dispatcher_calls::gate): Likewise.
28855 (make_pass_dispatcher_calls): Likewise.
28856 * passes.def (pass_target_clone): Put as very first IPA early
28859 2017-02-03 Martin Liska <mliska@suse.cz>
28861 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
28862 in case of a function with ifunc attribute.
28864 2017-02-03 Martin Liska <mliska@suse.cz>
28866 * cgraph.c (cgraph_node::dump): Dump function version info.
28867 * symtab.c (symtab_node::dump_base): Add missing new line.
28869 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28871 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
28872 (ifcombine_ifandif): Use it.
28874 2017-02-03 Martin Liska <mliska@suse.cz>
28876 * doc/invoke.texi: Document default value for
28877 use-after-scope-direct-emission-threshold.
28879 2017-02-03 Martin Liska <mliska@suse.cz>
28881 PR tree-optimization/79339
28882 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
28883 (format_floating): Likewise.
28885 2017-02-03 Martin Liska <mliska@suse.cz>
28888 * ipa-prop.c (ipa_node_params_t::insert): Remove current
28890 (ipa_node_params_t::remove): Likewise.
28891 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
28892 initialization from removed ipa_node_params_t::insert.
28893 (ipa_node_params::~ipa_node_params): Move from removed
28894 ipa_node_params_t::release.
28895 * symbol-summary.h (symbol_summary::m_released): New member.
28896 Do not release a summary twice. Do not allow to call finalizer
28897 for types of a summary that live in GGC memory.
28899 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28901 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
28904 2017-02-02 Martin Sebor <msebor@redhat.com>
28906 PR middle-end/79275
28907 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
28908 (format_string): Tighten up the range of output for non-constant
28909 strings and correct the expected range for wide non-constant strings.
28911 2017-02-02 Martin Sebor <msebor@redhat.com>
28913 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
28915 PR middle-end/32003
28916 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
28918 (-fdump-tree-@var): Add to index and document how to come up
28919 with pass-specific option and dump file names.
28920 (-fdump-passes): Clarify where to look for output.
28922 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28924 PR middle-end/77445
28925 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
28926 statistics of the analyzed path; allow threading for speed when
28927 any of BBs along the path are optimized for speed.
28929 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
28931 PR middle-end/78468
28932 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
28933 settings of the virtual registers.
28936 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
28938 * explow.c (get_dynamic_stack_size): Take known alignment of stack
28939 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
28942 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28944 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
28945 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
28947 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28949 * config/s390/s390.md: Add missing comments with the expanded
28951 * config/s390/vector.md: Likewise.
28952 * config/s390/vx-builtins.md: Likewise.
28954 2017-02-02 Jakub Jelinek <jakub@redhat.com>
28957 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
28958 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
28959 conditions on a single line.
28961 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28963 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
28964 __S390_VX__ to __VX__.
28966 2017-02-01 Andrew Pinski <apinski@cavium.com>
28968 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
28969 stmt_info to record_stmt_cost.
28970 (vect_get_known_peeling_cost): Pass stmt_info if known to
28972 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
28973 cpu_vector_cost field into
28974 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
28975 field into vec_int_stmt_cost and vec_fp_stmt_cost.
28976 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
28977 splitting of scalar_stmt_cost and vec_stmt_cost.
28978 (thunderx_vector_cost): Likewise.
28979 (cortexa57_vector_cost): LIkewise.
28980 (exynosm1_vector_cost): Likewise.
28981 (xgene1_vector_cost): Likewise.
28982 (thunderx2t99_vector_cost): Improve after the splitting of the two
28984 (aarch64_builtin_vectorization_cost): Update for the splitting of
28985 scalar_stmt_cost and vec_stmt_cost.
28987 2017-02-01 Torvald Riegel <triegel@redhat.com>
28988 Richard Henderson <rth@redhat.com>
28990 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
28991 conditional on existance of a fast atomic load.
28992 * optabs-query.c (can_atomic_load_p): New function.
28993 * optabs-query.h (can_atomic_load_p): Declare it.
28994 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
28995 no fast atomic load is available for the particular size of access.
28996 (expand_atomic_compare_and_swap): Likewise.
28997 (expand_atomic_load): Likewise.
28998 (expand_atomic_store): Likewise.
28999 (expand_atomic_fetch_op): Likewise.
29000 * testsuite/lib/target-supports.exp
29001 (check_effective_target_sync_int_128): Remove x86 because it provides
29002 no fast atomic load.
29003 (check_effective_target_sync_int_128_runtime): Likewise.
29005 2017-02-01 Richard Biener <rguenther@suse.de>
29007 * graphite.c: Include tree-vectorizer.h for find_loop_location.
29008 (graphite_transform_loops): Provide opt-info for optimized nests.
29009 * tree-parloop.c (parallelize_loops): Provide opt-info for
29010 parallelized loops.
29012 2017-02-01 Richard Biener <rguenther@suse.de>
29014 PR middle-end/79315
29015 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
29016 was not set before.
29018 2017-02-01 Richard Biener <rguenther@suse.de>
29020 PR tree-optimization/71824
29021 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29022 Verify the loops are valid in the merged SESE region.
29023 (scop_detection::can_represent_loop_1): Check analyzing the
29024 evolution of the number of iterations in the region succeeds.
29026 2017-01-31 Ian Lance Taylor <iant@golang.org>
29028 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
29029 REG_ARGS_SIZE note to 32-bit push insns and call insn.
29031 2017-01-31 David Malcolm <dmalcolm@redhat.com>
29033 PR preprocessor/79210
29034 * input.c (get_substring_ranges_for_loc): Replace line_width
29035 assertion with error-handling.
29037 2017-01-31 Richard Biener <rguenther@suse.de>
29039 PR tree-optimization/77318
29040 * graphite-sese-to-poly.c (extract_affine): Fix assert.
29041 (create_pw_aff_from_tree): Take loop parameter.
29042 (add_condition_to_pbb): Pass loop of the condition to
29043 create_pw_aff_from_tree.
29045 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29047 * config/s390/s390.c (s390_asan_shadow_offset): New function.
29048 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
29050 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
29054 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
29056 (convert_int_to_float128): Likewise.
29057 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
29058 (convert_int_to_float128): Likewise.
29059 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
29060 (UNSPEC_IEEE128_CONVERT): Likewise.
29061 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
29062 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
29063 Use local variables for IBM extended format.
29064 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
29065 (fix_trunc<mode>si2_fprs): Likewise.
29066 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
29067 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
29068 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
29069 to know that we can now have integers of all sizes in vector
29071 (fix<uns>_<mode>di2_hw): Likewise.
29072 (float<uns>_<mode>si2_hw): Likewise.
29073 (fix_<mode>si2_hw): Likewise.
29074 (fixuns_<mode>si2_hw): Likewise.
29075 (float<uns>_<mode>di2_hw): Likewise.
29076 (float_<mode>di2_hw): Likewise.
29077 (float_<mode>si2_hw): Likewise.
29078 (floatuns_<mode>di2_hw): Likewise.
29079 (floatuns_<mode>si2_hw): Likewise.
29080 (xscvqp<su>wz_<mode>): Delete, no longer used.
29081 (xscvqp<su>dz_<mode>): Likewise.
29082 (xscv<su>dqp_<mode>): Likewise.
29083 (ieee128_mfvsrd_64bit): Likewise.
29084 (ieee128_mfvsrd_32bit): Likewise.
29085 (ieee128_mfvsrwz): Likewise.
29086 (ieee128_mtvsrw): Likewise.
29087 (ieee128_mtvsrd_64bit): Likewise.
29088 (ieee128_mtvsrd_32bit): Likewise.
29090 2017-01-31 Martin Liska <mliska@suse.cz>
29093 * ipa-prop.c (ipa_free_all_node_params): Call release method
29094 instead of ~sumbol_summary to not to trigger double times
29097 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
29099 PR tree-optimization/71691
29100 * bitmap.h (class auto_bitmap): New.
29101 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
29102 is_maybe_undefined instead of ssa_undefined_value_p.
29104 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29106 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
29107 __S390_ARCH_LEVEL__ to __ARCH__.
29109 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29111 PR tree-optimization/79267
29112 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
29113 if should_remove_lhs_p is true.
29115 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
29118 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
29119 (add_alignment_attribute): New.
29120 (base_type_die): Add alignment attribute.
29121 (subrange_type_die): Likewise.
29122 (modified_type_die): Likewise.
29123 (gen_array_type_die): Likewise.
29124 (gen_descr_array_type_die: Likewise.
29125 (gen_enumeration_type_die): Likewise.
29126 (gen_subprogram_die): Likewise.
29127 (gen_variable_die): Likewise.
29128 (gen_field_die): Likewise.
29129 (gen_ptr_to_mbr_type_die): Likewise.
29130 (gen_struct_or_union_type_die): Likewise.
29131 (gen_subroutine_type_die): Likewise.
29132 (gen_typedef_die): Likewise.
29133 (base_type_cmp): Compare alignment attribute.
29135 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29138 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
29139 (setb_unsigned) New pattern for setb with CCUNS.
29140 * config/rs6000/rs6000.c (expand_block_compare): Use a different
29141 subfc./subfe sequence to avoid overflow problems. Generate a
29142 shorter sequence with cmpld/setb for power9.
29143 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
29144 for generating subfc. instruction.
29145 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
29146 now uses this instruction.
29148 2017-01-30 Ian Lance Taylor <iant@google.com>
29151 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
29152 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
29154 2017-01-30 Martin Sebor <msebor@redhat.com>
29156 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
29157 Move constant to the right of a relational operator.
29158 (get_mpfr_format_length, format_character, format_string): Ditto.
29159 (should_warn_p, maybe_warn): Same.
29161 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
29163 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
29166 * asan.c (get_translation_unit_decl): Remove function.
29167 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
29169 2017-01-30 Martin Liska <mliska@suse.cz>
29171 PR gcov-profile/79259
29172 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
29173 -fprofile-generate.
29175 2017-01-30 Martin Liska <mliska@suse.cz>
29178 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
29179 Initialize variables with NULL value.
29181 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
29184 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
29186 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
29188 2017-01-30 Richard Biener <rguenther@suse.de>
29190 PR tree-optimization/79276
29191 * tree-vrp.c (process_assert_insertions): Properly adjust common
29192 when removing a duplicate.
29194 * gcc.dg/torture/pr79276.c: New testcase.
29196 2017-01-30 Richard Biener <rguenther@suse.de>
29198 PR tree-optimization/79256
29199 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
29200 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
29202 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
29204 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
29207 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
29208 ("*r<noxa>sbg_<mode>_sll_bitmask")
29209 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
29210 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
29211 Use contiguous_bitmask_nowrap_operand.
29213 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29216 * config/rs6000/altivec.h (vec_xl): Revise #define.
29217 (vec_xst): Likewise.
29219 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
29221 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
29223 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
29225 PR rtl-optimization/79194
29226 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
29227 traps before call to bypass_conditional_jumps.
29229 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
29231 PR tree-optimization/71374
29232 * lra-constraints.c (check_conflict_input_operands): New.
29233 (match_reload): Use it.
29235 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
29238 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
29239 account to calculate conflict_set.
29241 2017-01-27 Bin Cheng <bin.cheng@arm.com>
29243 PR rtl-optimization/78559
29244 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
29245 other_insn in combine.
29247 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29249 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
29250 uint16_type_node for BT_UINT16.
29252 2017-01-27 David Malcolm <dmalcolm@redhat.com>
29254 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
29255 "RTL Tests" to menu.
29256 (GIMPLE Tests): New node.
29257 (RTL Tests): New node.
29259 2017-01-27 Richard Biener <rguenther@suse.de>
29261 PR tree-optimization/79245
29262 * tree-loop-distribution.c (distribute_loop): Apply cost
29263 modeling also to detected patterns.
29265 2017-01-27 Richard Biener <rguenther@suse.de>
29267 PR tree-optimization/71433
29268 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
29269 (compare_assert_loc): New function.
29270 (process_assert_insertions): Sort and optimize assert locations
29271 to remove duplicates and push down identical assertions on
29272 edges to their destination block.
29274 2017-01-27 Richard Biener <rguenther@suse.de>
29276 PR tree-optimization/79244
29277 * tree-vrp.c (remove_range_assertions): Forcefully propagate
29278 out SSA names even if abnormal.
29280 2017-01-27 Jakub Jelinek <jakub@redhat.com>
29282 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
29283 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
29284 instead of MPFR_RNDN.
29286 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
29289 * arm.c (arm_option_override): Don't call build_target_option_node
29290 until after doing all option overrides.
29291 (arm_valid_target_attribute_tree): Likewise.
29293 2017-01-27 Martin Liska <mliska@suse.cz>
29295 * doc/invoke.texi (-fprofile-arcs): Document profiling support
29296 for {cd}tors and C++ {cd}tors.
29298 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
29300 * config/s390/s390.md ("*setmem_long_and")
29301 ("*setmem_long_and_31z"): Use zero_extend instead of and.
29303 2017-01-26 Martin Sebor <msebor@redhat.com>
29305 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
29308 2017-01-26 Martin Sebor <msebor@redhat.com>
29310 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
29311 HAVE_DFmode before using XFmode or DFmode.
29312 (parse_directive): Avoid using the z length modifier to avoid
29313 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
29315 PR middle-end/78703
29316 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
29317 to accept adjustment as an array.
29318 (get_int_range): New function.
29319 (struct directive): Make width and prec arrays.
29320 (directive::set_width, directive::set_precision): Call get_int_range.
29321 (format_integer, format_floating): Handle width and precision ranges.
29322 (format_string, parse_directive): Same.
29324 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29327 * dwarf2out.c (generate_skeleton_bottom_up): For children with
29328 comdat_type_p set, just clone them, but keep the children in the
29332 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
29333 which have direct callers with -fvar-tracking-assignments enabled
29335 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
29336 inside of type units.
29338 2017-01-26 Martin Sebor <msebor@redhat.com>
29340 PR middle-end/78703
29341 * gimple-ssa-sprintf.c (struct result_range): Add likely and
29343 (struct format_result): Replace number_chars, number_chars_min,
29344 and number_chars_max with a single member of struct result_range.
29346 (format_result::operator+=): Adjust.
29347 (struct fmtresult): Remove bounded. Handle likely and unlikely
29349 (fmtresult::adjust_for_width_or_precision): New function.
29350 (fmtresult:type_max_digits): New function.
29351 (bytes_remaining): Handle likely and unlikely counters.
29352 (min_bytes_remaining): Remove.
29353 (format_percent): Simplify.
29354 (format_integer, format_floating): Set likely and unlikely counters.
29355 (get_string_length, format_character, format_string): Same.
29356 (format_plain, should_warn_p): New function.
29357 (maybe_warn): Call should_warn_p. Update diagnostic messages
29358 and handle those for all directives, including plain strings.
29359 (format_directive): Handle likely and unlikely counters.
29360 Remove unnecessary quoting from diagnostics. Add an informational
29362 (add_bytes): Remove.
29363 (pass_sprintf_length::compute_format_length): Simplify.
29364 (try_substitute_return_value): Handle likely and unlikely counters.
29366 2017-01-26 Carl Love <cel@us.ibm.com>
29368 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
29369 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
29371 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
29374 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
29375 endianess for subregs into account.
29376 * lra-constraints.c (lra_constraints): Do risky transformations
29377 always on the first iteration.
29378 * lra-lives.c (check_pseudos_live_through_calls): Add arg
29379 last_call_used_reg_set.
29380 (process_bb_lives): Define and use last_call_used_reg_set.
29381 * lra.c (lra): Always continue after lra_constraints on the first
29384 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
29386 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
29388 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
29390 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29392 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
29393 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
29394 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
29395 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
29396 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
29397 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
29398 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
29399 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
29400 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
29402 2017-01-26 Marek Polacek <polacek@redhat.com>
29405 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
29406 for the third operand.
29408 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29410 PR middle-end/79236
29411 * omp-low.c (struct omp_context): Add simt_stmt field.
29412 (scan_omp_for): Return omp_context *.
29413 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
29414 context to the _simt_ SIMD stmt.
29415 (lower_omp_for): For combined SIMD with sibling _simt_
29416 SIMD, make sure to use the same decls in _looptemp_
29417 clauses as in the sibling.
29419 2017-01-26 David Sherwood <david.sherwood@arm.com>
29421 PR middle-end/79212
29422 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
29425 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29428 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
29429 emit fld b; fld a; if possible.
29431 * brig-builtins.def: Update copyright years.
29432 * config/arm/arm_acle_builtins.def: Update copyright years.
29434 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
29437 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
29438 constraint instead of o for the stxsd instruction.
29440 2017-01-25 Carl Love <cel@us.ibm.com>
29442 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
29443 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
29445 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
29447 * doc/invoke.texi (C++ Dialect Options): Fix typo.
29449 2017-01-25 Richard Biener <rguenther@suse.de>
29451 PR tree-optimization/69264
29452 * target.def (vector_alignment_reachable): Improve documentation.
29453 * doc/tm.texi: Regenerate.
29454 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
29456 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
29457 earlier changes with respect to TYPE_USER_ALIGN.
29458 (vector_alignment_reachable_p): Likewise. Improve dumping.
29460 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29463 * config/arm/arm.md (xordi3): Force constant operand into a register
29466 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29468 * doc/invoke.texi (-fstore-merging): Correct default optimization
29469 levels at which it is enabled.
29470 (-O): Move -fstore-merging from list to...
29473 2017-01-25 Richard Biener <rguenther@suse.de>
29476 * omp-expand.c: Include debug.h.
29477 (expand_omp_taskreg): Make sure to generate early debug before
29478 outlining anything from a function.
29479 (expand_omp_target): Likewise.
29480 (grid_expand_target_grid_body): Likewise.
29482 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
29485 * asan.c (get_translation_unit_decl): New function.
29486 (asan_add_global): Extract modules file name from globals
29487 TRANSLATION_UNIT_DECL name.
29489 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
29492 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
29493 for long calls with APCS frame and VFP.
29495 2017-01-24 David Malcolm <dmalcolm@redhat.com>
29497 * cfg.c (original_copy_tables_initialized_p): New function.
29498 * cfg.h (original_copy_tables_initialized_p): New decl.
29499 * cfgrtl.c (relink_block_chain): Guard the call to
29500 free_original_copy_tables with a call to
29501 original_copy_tables_initialized_p.
29502 * cgraph.h (symtab_node::native_rtl_p): New decl.
29503 * cgraphunit.c (symtab_node::native_rtl_p): New function.
29504 (symtab_node::needed_p): Don't assert for early assembly output
29505 for __RTL functions.
29506 (cgraph_node::finalize_function): Set "force_output" for __RTL
29508 (cgraph_node::analyze): Bail out early for __RTL functions.
29509 (analyze_functions): Update assertion to support __RTL functions.
29510 (cgraph_node::expand): Bail out early for __RTL functions.
29511 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
29513 * function.h (struct function): Update comment for field
29515 * gimple-expr.c: Include "tree-pass.h".
29516 (gimple_has_body_p): Return false for __RTL functions.
29517 * Makefile.in (OBJS): Add run-rtl-passes.o.
29518 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
29520 (gcc::pass_manager::get_clean_slate): New accessor.
29521 * passes.c: Include "insn-addr.h".
29522 (should_skip_pass_p): Add logging. Update logic for running
29523 "expand" to be compatible with both __GIMPLE and __RTL. Guard
29524 property-provider override so it is only done for gimple passes.
29526 (skip_pass): New function.
29527 (execute_one_pass): Call skip_pass when skipping passes.
29528 * read-md.c (md_reader::read_char): Support filtering
29529 the input to a subset of line numbers.
29530 (md_reader::md_reader): Initialize fields
29531 m_first_line and m_last_line.
29532 (md_reader::read_file_fragment): New function.
29533 * read-md.h (md_reader::read_file_fragment): New decl.
29534 (md_reader::m_first_line): New field.
29535 (md_reader::m_last_line): New field.
29536 * read-rtl-function.c (function_reader::create_function): Only
29537 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
29538 curr_properties. Set DECL_INITIAL to a dummy block.
29539 (read_rtl_function_body_from_file_range): New function.
29540 * read-rtl-function.h (read_rtl_function_body_from_file_range):
29542 * run-rtl-passes.c: New file.
29543 * run-rtl-passes.h: New file.
29545 2017-01-24 Jeff Law <law@redhat.com>
29547 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
29550 2017-01-24 Bin Cheng <bin.cheng@arm.com>
29552 PR tree-optimization/79159
29553 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
29554 (record_nonwrapping_iv): Improve boundary using above function if no
29555 value range information.
29557 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
29558 Martin Jambor <mjambor@suse.cz>
29560 * brig-builtins.def: New file.
29561 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
29562 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
29563 (DEF_HSAIL_SAT_BUILTIN): Likewise.
29564 (DEF_HSAIL_INTR_BUILTIN): Likewise.
29565 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
29566 * builtin-types.def (BT_INT8): New.
29567 (BT_INT16): Likewise.
29568 (BT_UINT8): Likewise.
29569 (BT_UINT16): Likewise.
29570 (BT_FN_ULONG): Likewise.
29571 (BT_FN_UINT_INT): Likewise.
29572 (BT_FN_UINT_ULONG): Likewise.
29573 (BT_FN_UINT_LONG): Likewise.
29574 (BT_FN_UINT_PTR): Likewise.
29575 (BT_FN_ULONG_PTR): Likewise.
29576 (BT_FN_INT8_FLOAT): Likewise.
29577 (BT_FN_INT16_FLOAT): Likewise.
29578 (BT_FN_UINT32_FLOAT): Likewise.
29579 (BT_FN_UINT16_FLOAT): Likewise.
29580 (BT_FN_UINT8_FLOAT): Likewise.
29581 (BT_FN_UINT64_FLOAT): Likewise.
29582 (BT_FN_UINT16_UINT32): Likewise.
29583 (BT_FN_UINT32_UINT16): Likewise.
29584 (BT_FN_UINT16_UINT16_UINT16): Likewise.
29585 (BT_FN_INT_PTR_INT): Likewise.
29586 (BT_FN_UINT_PTR_UINT): Likewise.
29587 (BT_FN_LONG_PTR_LONG): Likewise.
29588 (BT_FN_ULONG_PTR_ULONG): Likewise.
29589 (BT_FN_VOID_UINT64_UINT64): Likewise.
29590 (BT_FN_UINT8_UINT8_UINT8): Likewise.
29591 (BT_FN_INT8_INT8_INT8): Likewise.
29592 (BT_FN_INT16_INT16_INT16): Likewise.
29593 (BT_FN_INT_INT_INT): Likewise.
29594 (BT_FN_UINT_FLOAT_UINT): Likewise.
29595 (BT_FN_FLOAT_UINT_UINT): Likewise.
29596 (BT_FN_ULONG_UINT_UINT): Likewise.
29597 (BT_FN_ULONG_UINT_PTR): Likewise.
29598 (BT_FN_ULONG_ULONG_ULONG): Likewise.
29599 (BT_FN_UINT_UINT_UINT): Likewise.
29600 (BT_FN_VOID_UINT_PTR): Likewise.
29601 (BT_FN_UINT_UINT_PTR: Likewise.
29602 (BT_FN_UINT32_UINT64_PTR): Likewise.
29603 (BT_FN_INT_INT_UINT_UINT): Likewise.
29604 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
29605 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
29606 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
29607 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
29608 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
29609 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
29610 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
29611 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
29612 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
29613 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
29614 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
29615 * doc/frontends.texi: List BRIG FE.
29616 * doc/install.texi (Testing): Add BRIG tesring requirements.
29617 * doc/invoke.texi (Overall Options): Mention BRIG.
29618 * doc/standards.texi (Standards): Doucment BRIG HSA version.
29620 2017-01-24 Richard Biener <rguenther@suse.de>
29622 PR translation/79208
29623 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
29625 2017-01-24 Martin Jambor <mjambor@suse.cz>
29628 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
29629 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
29630 and known_contexts.
29632 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
29634 PR middle-end/79123
29635 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
29636 casts from signed to unsigned really don't have a range.
29638 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
29640 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
29641 GMP_RNDx for compatiblity.
29643 2017-01-24 Martin Liska <mliska@suse.cz>
29646 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
29647 that would prevent us to call alloca with -1 as argument.
29649 2017-01-24 Jakub Jelinek <jakub@redhat.com>
29651 * dwarf2out.c (output_compilation_unit_header, output_file_names):
29652 Avoid -Wformat-security warning.
29654 2017-01-23 Andrew Pinski <apinski@cavium.com>
29656 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
29659 2017-01-23 Martin Sebor <msebor@redhat.com>
29661 PR middle-end/78703
29662 * gimple-ssa-sprintf.c (warn_level): New global.
29663 (format_integer): Use it here and throughout the rest of the file.
29664 Use the same switch to compute sign as base.
29665 (maybe_warn): New function.
29666 (format_directive): Factor out warnings into maybe_warn.
29667 Add debugging output. Use warn_level.
29668 (add_bytes): Use warn_level.
29669 (pass_sprintf_length::compute_format_length): Add debugging output.
29670 (try_substitute_return_value): Same.
29671 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
29673 PR middle-end/78703
29674 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
29675 (struct fmtresult, format_integer, format_floating): Adjust.
29676 (fmtresult::fmtresult): Set max correctly in two argument ctor.
29677 (get_string_length, format_string,format_directive): Same.
29678 (pass_sprintf_length::compute_format_length): Same.
29679 (try_substitute_return_value): Simplify slightly.
29681 PR middle-end/78703
29682 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
29683 (fmtresult::operator+=): Outlined.
29684 (struct fmtresult): Add ctors.
29685 (struct conversion_spec): Rename...
29686 (struct directive): ...to this. Add and remove data members.
29687 (directive::set_width, directive::set_precision): New functions.
29688 (format_percent): Use fmtresult ctor.
29689 (get_width_and_precision): Remove.
29690 (format_integer): Make naming changes. Avoid computing width and
29692 (format_floating): Same. Adjust indentation.
29693 (format_character, format_none): New functions.
29694 (format_string): Moved character handling to format_character.
29695 (format_directive): Remove arguments, change return type.
29696 (parse_directive): New function.
29697 (pass_sprintf_length::compute_format_length): Move directive
29698 parsing to parse_directive.
29700 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29702 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
29703 (assign_assembler_name_if_needed): ... this.
29704 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
29705 (assign_assembler_name_if_needed): ... this.
29706 (free_lang_data_in_cgraph): Adjust callers.
29707 * cgraphunit.c (cgraph_node::analyze): Likewise.
29708 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
29711 2017-01-23 Richard Biener <rguenther@suse.de>
29713 PR tree-optimization/79088
29714 PR tree-optimization/79188
29715 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
29716 resetting loop bounds after last path deletion. Reset loop
29717 bounds of the target loop, make code match the comments.
29718 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
29719 Make sure loops need no fixups.
29721 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29723 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
29724 exponent support with double type for first argument.
29725 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
29726 type returned by __builtin_vec_extract_sig,
29727 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
29728 functions from "vector int" to "vector unsigned int" or from
29729 "vector long long int" to "vector unsigned long long int".
29730 Changed type returned by __builtin_vec_extract_exp,
29731 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
29732 functions from "vector int" to "vector unsigned int" or from
29733 "vector long long int" to "vector unsigned long long int".
29734 Changed return type of __builtin_vec_test_data_class,
29735 __builtin_vec_test_data_class_sp, and
29736 __builtin_vec_test_data_class_dp from "vector int" to
29737 "vector bool int" or from "vector long long int" to "vector bool
29738 long long int" and changed second argument type from "unsigned
29739 int" to "int". Added new overloaded function forms "vector float
29740 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
29741 "vector float __builtin_vec_insert_exp_sp (vector float, vector
29742 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
29743 double, vector unsigned long long int)" and "vector double
29744 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
29745 long int)". Changed return type of
29746 __builtin_scalar_test_data_class and
29747 __builtin_scalar_test_data_class_sp and
29748 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
29749 int" and changed second argument from "unsigned int" to "int".
29750 Changed type returned by __builtin_scalar_test_neg,
29751 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
29752 from "int" to "bool int". Added new overloaded function form
29753 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
29754 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
29755 exponent double-precision with floating point first argument.
29756 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
29757 documentation of scalar_test_data_class, scalar_test_neg,
29758 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
29759 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
29760 vec_test_data_class built-in functions to reflect refinements in
29761 their type signatures.
29763 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
29765 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
29767 (aarch64_elf_asm_destructor): Likewise.
29769 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
29771 PR rtl-optimization/78634
29772 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
29773 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
29774 * ifcvt.c (noce_try_cmove): Add missing cost check.
29776 PR rtl-optimization/71724
29777 * combine.c (if_then_else_cond): Look for situations where it is
29778 beneficial to undo the work of one of the recursive calls.
29780 2017-01-23 Bin Cheng <bin.cheng@arm.com>
29782 PR tree-optimization/70754
29783 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
29784 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
29785 combined stmt before it if not NULL.
29786 (combine_chains): Process refs reversely and compute dominance point
29789 2017-01-23 Martin Liska <mliska@suse.cz>
29791 PR tree-optimization/79196
29792 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
29793 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
29795 (strlen_optimize_stmt): Call the renamed function.
29797 2017-01-23 Michael Matz <matz@suse.de>
29799 PR tree-optimization/78384
29800 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
29802 2017-01-23 Richard Biener <rguenther@suse.de>
29804 PR tree-optimization/79186
29805 * tree-vrp.c (register_new_assert_for): Make sure we've seen
29806 both incoming edges before moving an assert.
29808 2017-01-23 Martin Jambor <mjambor@suse.cz>
29810 * ipa-prop.c (load_from_param_1): Removed.
29811 (load_from_unmodified_param): Bits from load_from_param_1 put back
29813 (load_from_param): Removed.
29814 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
29815 with stmt. Reverted back to use of load_from_unmodified_param.
29817 2017-01-23 Martin Jambor <mjambor@suse.cz>
29820 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
29821 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
29822 field a pointer to garbage collected vector, mark lattices and
29823 ipcp_orig_node with GTY((skip)).
29824 (ipa_get_param_count): Adjust to descriptors being a pointer.
29825 (ipa_get_param): Likewise.
29826 (ipa_get_type): Likewise.
29827 (ipa_get_param_move_cost): Likewise.
29828 (ipa_set_param_used): Likewise.
29829 (ipa_get_controlled_uses): Likewise.
29830 (ipa_set_controlled_uses): Likewise.
29831 (ipa_is_param_used): Likewise.
29832 (ipa_node_params_t): Move into garbage collector. New methods insert
29834 (ipa_node_params_sum): Annotate wth GTY(()).
29835 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
29837 (ipa_load_from_parm_agg): Adjust declaration.
29838 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
29839 * ipa-profile.c (ipa_profile): Likewise.
29840 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
29841 (ipa_populate_param_decls): Make descriptors parameter garbage
29843 (ipa_dump_param): Adjust to descriptors being a pointer.
29844 (ipa_alloc_node_params): Likewise.
29845 (ipa_initialize_node_params): Likewise.
29846 (load_from_param_1): Make descriptors parameter garbage collected.
29847 (load_from_unmodified_param): Likewise.
29848 (load_from_param): Likewise.
29849 (ipa_load_from_parm_agg): Likewise.
29850 (ipa_node_params::~ipa_node_params): Removed.
29851 (ipa_free_all_node_params): Remove call to delete operator.
29852 (ipa_node_params_t::insert): New.
29853 (ipa_node_params_t::remove): Likewise.
29854 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
29855 copy known_csts and known_contexts vectors.
29856 (ipa_read_node_info): Adjust to descriptors being a pointer.
29857 (ipcp_modif_dom_walker): Make m_descriptors field garbage
29859 (ipcp_transform_function): Make descriptors variable garbage
29862 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
29864 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
29865 * config/i386/avx512dqintrin.h: Ditto.
29866 * config/i386/avx512fintrin.h: Ditto.
29867 * config/i386/i386.c: Handle new builtins.
29868 * config/i386/i386-builtin.def: Add new builtins.
29869 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
29870 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
29872 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29873 Martin Liska <mliska@suse.cz>
29875 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
29876 * asan.c (asan_expand_poison_ifn): Support stores and use
29877 appropriate ASAN report function.
29878 * internal-fn.c (expand_ASAN_POISON_USE): New function.
29879 * internal-fn.def (ASAN_POISON_USE): Declare.
29880 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
29881 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
29882 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
29883 ASAN_POISON calls w/o LHS.
29884 * tree-ssa.c (execute_update_addresses_taken): Create clobber
29885 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
29886 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
29887 * gimplify.c (asan_poison_variables): Add attribute
29888 use_after_scope_memory to variables that really needs to live
29890 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
29891 having the attribute.
29893 2017-01-23 Martin Liska <mliska@suse.cz>
29895 * asan.c (create_asan_shadow_var): New function.
29896 (asan_expand_poison_ifn): Likewise.
29897 * asan.h (asan_expand_poison_ifn): New declaration.
29898 * internal-fn.c (expand_ASAN_POISON): Likewise.
29899 * internal-fn.def (ASAN_POISON): New builtin.
29900 * sanopt.c (pass_sanopt::execute): Expand
29901 asan_expand_poison_ifn.
29902 * tree-inline.c (copy_decl_for_dup_finish): Make function
29904 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
29905 * tree-ssa.c (is_asan_mark_p): New function.
29906 (execute_update_addresses_taken): Rewrite local variables
29907 (identified just by use-after-scope as addressable) into SSA.
29909 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
29911 * doc/install.texi (Specific): opensource.apple.com uses https
29912 now. Remove trailing slash.
29914 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
29916 * README.Portability: Remove note on an Irix compatibility issue.
29918 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
29920 * gcov.c (INCLUDE_ALGORITHM): Define.
29921 (INCLUDE_VECTOR): Define.
29922 No longer include <vector> and <algorithm> directly.
29924 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
29926 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
29928 * doc/invoke.texi (Code Gen Options): Ditto.
29930 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
29933 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
29935 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
29937 rtl-optimization/79125
29938 * cprop.c (local_cprop_pass): Handle cases where we make an
29939 unconditional trap.
29941 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
29945 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
29946 read from, for big endian.
29948 2017-01-20 Jiong Wang <jiong.wang@arm.com>
29950 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
29951 register pauth builtins for LP64 only.
29953 2017-01-20 Marek Polacek <polacek@redhat.com>
29956 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
29959 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
29961 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
29963 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
29964 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
29965 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
29967 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29970 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
29971 in big-endian mode when they are not a single duplicated value.
29973 2017-01-20 Richard Biener <rguenther@suse.de>
29975 * BASE-VER: Bump to 7.0.1.
29977 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
29979 * omp-low.c (omplow_simd_context): New struct. Use it...
29980 (lower_rec_simd_input_clauses): ...here and...
29981 (lower_rec_input_clauses): ...here to hold common data. Adjust all
29982 references to idx, lane, max_vf, is_simt.
29984 2017-01-20 Graham Markall <graham.markall@embecosm.com>
29986 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
29989 2017-01-20 Martin Jambor <mjambor@suse.cz>
29991 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
29992 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
29994 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
29995 (GTFILES): Rename hsa.c to hsa-common.c.
29996 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
29997 * hsa-dump.c: Likewise.
29998 * hsa-gen.c: Likewise.
29999 * hsa-regalloc.c: Likewise.
30000 * ipa-hsa.c: Likewise.
30001 * omp-expand.c: Likewise.
30002 * omp-low.c: Likewise.
30003 * toplev.c: Likewise.
30005 2017-01-20 Marek Polacek <polacek@redhat.com>
30008 * doc/invoke.texi: Document -Wduplicated-branches.
30009 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
30010 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
30011 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
30012 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
30013 return 0 only when not OEP_LEXICOGRAPHIC.
30014 (fold_build_cleanup_point_expr): Use the expression
30015 location when building CLEANUP_POINT_EXPR.
30016 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
30017 * tree.c (add_expr): Handle error_mark_node.
30019 2017-01-20 Martin Liska <mliska@suse.cz>
30022 * tree-profile.c (init_ic_make_global_vars): Do not call
30024 (gimple_init_gcov_profiler): Likewise.
30026 2017-01-20 Martin Liska <mliska@suse.cz>
30029 * cgraph.h (maybe_create_reference): Remove argument and
30031 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
30033 * ipa-cp.c (create_specialized_node): Likewise.
30034 * symtab.c (symtab_node::maybe_create_reference): Handle
30035 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
30037 2017-01-20 Martin Liska <mliska@suse.cz>
30039 * read-rtl-function.c (function_reader::create_function): Use
30040 build_decl instread of build_decl_stat.
30042 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
30044 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
30045 * config/i386/avx512dqintrin.h: Ditto.
30046 * config/i386/avx512fintrin.h: Ditto.
30047 * config/i386/i386-builtin-types.def: Add new types.
30048 * config/i386/i386.c: Handle new types.
30049 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
30050 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
30051 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
30052 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
30053 (__builtin_ia32_kshiftridi): New.
30054 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
30056 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
30060 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
30061 define to rs6000_init_stack_protect_guard.
30062 (rs6000_init_stack_protect_guard): New function.
30064 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30065 Yunqiang Su <yunqiang.su@imgtec.com>
30067 * config.gcc (supported_defaults): Add madd4.
30068 (with_madd4): Add validation.
30069 (all_defaults): Add madd4.
30070 * config/mips/mips.opt (mmadd4): New option.
30071 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
30073 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
30075 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
30076 (ISA_HAS_FUSED_MADD4): Likewise.
30077 * doc/invoke.texi (-mmadd4): Document the new option.
30078 * doc/install.texi (--with-madd4): Document the new option.
30080 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30082 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
30083 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
30084 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
30085 (aarch64_init_pauth_hint_builtins): New.
30086 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
30087 (aarch64_expand_builtin): Expand new builtins.
30089 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30091 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
30092 * combine-stack-adj.c (no_unhandled_cfa): Handle
30093 REG_CFA_TOGGLE_RA_MANGLE.
30094 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
30095 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
30096 info for return address signing.
30097 (aarch64_expand_epilogue): Likewise.
30099 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30101 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
30102 * config/aarch64/aarch64-protos.h
30103 (aarch64_return_address_signing_enabled): New declaration.
30104 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
30106 (aarch64_expand_prologue): Sign return address before it's pushed onto
30108 (aarch64_expand_epilogue): Authenticate return address fetched from
30110 (aarch64_override_options): Sanity check for ILP32 and ISA level.
30111 (aarch64_attributes): New function attributes for "sign-return-address".
30112 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
30113 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
30114 ("*do_return"): Generate combined instructions according to key index.
30115 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
30116 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
30118 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
30119 * config/aarch64/aarch64.opt (msign-return-address=): New.
30120 * doc/extend.texi (AArch64 Function Attributes): Documents
30121 "sign-return-address=".
30122 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
30124 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30126 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
30127 overall option summary.
30129 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30131 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
30132 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
30133 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
30134 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
30136 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
30138 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
30139 -mpower9-minmax by default for -mcpu=power9.
30140 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
30141 128-bit floating point.
30143 2017-01-20 Alan Modra <amodra@gmail.com>
30145 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
30146 optimizing for size.
30148 2017-01-20 Alan Modra <amodra@gmail.com>
30151 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
30152 for strcmp and strncmp from corresponding builtin decl.
30154 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
30156 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
30157 instead of i386/rtems-64.h.
30158 * config/i386/rtems-64.h: Remove.
30160 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
30164 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
30166 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
30168 2017-01-19 Tamar Christina <tamar.christina@arm.com>
30170 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
30171 Change int to HOST_WIDE_INT.
30172 * config/aarch64/aarch64-protos.h
30173 (aarch64_simd_gen_const_vector_dup): Likewise.
30174 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
30176 2017-01-19 David Malcolm <dmalcolm@redhat.com>
30178 * langhooks-def.h (lhd_type_for_size): New decl.
30179 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
30180 * langhooks.c (lhd_type_for_size): New function, taken from
30183 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
30185 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
30186 define_bypass for CR latency.
30187 (power9-cracked-alu): Update bypass latency and remove power9-branch.
30188 (power9-alu2): Add define_bypass for CR latency.
30190 (power9-mul): Update insn latency.
30191 (power9-mul-compare): Update insn latency, bypass latency and remove
30194 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30196 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
30198 * config/aarch64/aarch64.md
30199 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
30200 aarch64_nopcrelative_literal_loads.
30201 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
30203 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
30205 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
30206 TARGET_LOONGSON_3A.
30207 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
30209 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
30212 * config.gcc (supported_defaults): Add lxc1-sxc1.
30213 (with_lxc1_sxc1): Add validation.
30214 (all_defaults): Add lxc1-sxc1.
30215 * config/mips/mips.opt (mlxc1-sxc1): New option.
30216 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
30218 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
30219 __mips_no_lxc1_sxc1.
30220 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
30221 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
30222 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
30224 2017-01-19 Richard Biener <rguenther@suse.de>
30226 PR tree-optimization/72488
30227 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
30228 sure to restore SSA info.
30229 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
30231 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
30233 PR rtl-optimization/79121
30234 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
30235 of the inner type when shifting an extended value.
30237 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
30240 * symtab.c (symtab_node::equal_address_to): Fix comparing of
30241 interposable aliases.
30243 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
30246 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
30247 Use the evmergelohi instruction.
30248 (mov_si<mode>_e500_subreg4_2_le): Likewise.
30249 (mov_sitf_e500_subreg8_2_be): Likewise.
30250 (mov_sitf_e500_subreg12_2_le): Likewise.
30251 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
30252 (mov_si<mode>_e500_subreg4_2_be): Likewise.
30253 (mov_sitf_e500_subreg8_2_le): Likewise.
30254 (mov_sitf_e500_subreg12_2_be): Likewise.
30256 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30258 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
30259 attribute from vecsimple to vecperm.
30260 (altivec_vbpermq2): Likewise.
30262 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30265 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
30267 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30268 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
30269 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
30270 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
30271 case where N arg is SIZE_MAX.
30272 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
30273 (cmpstrsi): Add pattern.
30275 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30277 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30278 __builtin_vec_revb builtins.
30279 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
30280 built-in functions to support generation of the ISA 3.0 XXBR<x>
30281 vector byte reverse instructions.
30282 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
30283 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
30284 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
30285 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
30286 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
30287 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
30288 (P9V_BUILTIN_VEC_REVB): Likewise.
30289 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
30290 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
30291 (p9_xxbrq_v16qi): Likewise.
30292 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
30293 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
30294 (p9_xxbrh_v8hi): Likewise.
30295 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
30296 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
30297 vec_revb built-in functions.
30299 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
30301 PR rtl-optimization/78952
30302 * config/i386/i386.md (any_extract): New code iterator.
30303 (*insvqi_2): Use any_extract for source operand.
30304 (*insvqi_3): Use any_shiftrt for source operand.
30306 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
30308 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
30310 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
30312 2017-01-18 Matthias Klose <doko@ubuntu.com>
30314 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
30316 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30318 * config/rs6000/altivec.h (vec_bperm): Change #define.
30319 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
30320 (altivec_vbpermq2): New define_insn.
30321 (altivec_vbpermd): Likewise.
30322 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
30323 function interface.
30324 (VBPERMD): Likewise.
30325 (VBPERM): New polymorphic function interface.
30326 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
30327 Add entries for P9V_BUILTIN_VEC_VBPERM.
30328 * doc/extend.texi: Add interfaces for vec_bperm.
30330 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30332 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
30333 first letter of error messages.
30334 (s390_resolve_overloaded_builtin): Likewise.
30335 * config/s390/s390.c (s390_expand_builtin): Likewise.
30336 (s390_invalid_arg_for_unprototyped_fn): Likewise.
30337 (s390_valid_target_attribute_inner_p): Likewise.
30338 * config/s390/s390.md ("tabort"): Likewise.
30340 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
30342 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
30343 (ISA_AVOID_DIV_HILO): New macro.
30344 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
30345 (ISA_HAS_DDIV): Likewise.
30347 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30349 * doc/invoke.texi (fabi-version): Correct number of occurrences.
30351 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30353 * doc/invoke.texi (fabi-version): Spelling fix.
30355 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30358 * doc/invoke.texi (fabi-version): Mention mangling fix for
30361 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30364 * doc/invoke.texi (fabi-version): Document discriminator mangling.
30366 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
30369 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
30370 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
30372 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
30374 (stack_protect_test): Ditto.
30375 * config/rs6000/rs6000.opt (mstack-protector-guard=,
30376 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
30378 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
30379 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
30380 -mstack-protector-guard-offset=.
30381 (RS/6000 and PowerPC Options): Ditto.
30383 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30385 * config/i386/i386.h (MASK_CLASS_P): New define.
30386 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
30387 there are no registers from different register sets also when
30388 mask registers are used. Update function comment.
30389 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
30390 to (*k/*r) and (*k/*km) alternatives.
30392 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
30394 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
30395 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
30396 (EH_RETURN_HANDLER_RTX): New define.
30397 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
30398 Force frame pointer in EH return functions.
30399 (aarch64_expand_epilogue): Add barrier for eh_return.
30400 (aarch64_final_eh_return_addr): Remove.
30401 (aarch64_eh_return_handler_rtx): New function.
30402 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
30404 (aarch64_eh_return_handler_rtx): New prototype.
30406 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30408 * config/rs6000/altivec.h (vec_rlmi): New #define.
30409 (vec_vrlnm): Likewise.
30410 (vec_rlnm): Likewise.
30411 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
30412 (UNSPEC_VRLNM): Likewise.
30413 (VIlong): New mode iterator.
30414 (altivec_vrl<VI_char>mi): New define_insn.
30415 (altivec_vrl<VI_char>nm): Likewise.
30416 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
30418 (VRLDNM): Likewise.
30419 (RLNM): New polymorphic function entry.
30420 (VRLWMI): New monomorphic function entry.
30421 (VRLDMI): Likewise.
30422 (RLMI): New polymorphic function entry.
30423 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
30424 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
30425 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
30428 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30431 * dwarf2out.c (field_byte_offset): Restore the
30432 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
30433 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
30434 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
30437 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30440 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
30442 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30445 * dwarf2out.c (add_data_member_location_attribute): For constant
30446 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
30447 instead of DW_AT_data_member_location, DW_AT_bit_offset and
30448 DW_AT_byte_size attributes.
30450 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30452 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
30453 after forcing to constant memory when the code model is medium.
30455 2017-01-17 Julia Koval <julia.koval@intel.com>
30458 * config/i386/avx512fintrin.h
30459 (_mm512_i32gather_ps): Change __addr type to void const*.
30460 (_mm512_mask_i32gather_ps): Ditto.
30461 (_mm512_i32gather_pd): Ditto.
30462 (_mm512_mask_i32gather_pd): Ditto.
30463 (_mm512_i64gather_ps): Ditto.
30464 (_mm512_mask_i64gather_ps): Ditto.
30465 (_mm512_i64gather_pd): Ditto.
30466 (_mm512_mask_i64gather_pd): Ditto.
30467 (_mm512_i32gather_epi32): Ditto.
30468 (_mm512_mask_i32gather_epi32): Ditto.
30469 (_mm512_i32gather_epi64): Ditto.
30470 (_mm512_mask_i32gather_epi64): Ditto.
30471 (_mm512_i64gather_epi32): Ditto.
30472 (_mm512_mask_i64gather_epi32): Ditto.
30473 (_mm512_i64gather_epi64): Ditto.
30474 (_mm512_mask_i64gather_epi64): Ditto.
30475 (_mm512_i32scatter_ps): Change __addr type to void*.
30476 (_mm512_mask_i32scatter_ps): Ditto.
30477 (_mm512_i32scatter_pd): Ditto.
30478 (_mm512_mask_i32scatter_pd): Ditto.
30479 (_mm512_i64scatter_ps): Ditto.
30480 (_mm512_mask_i64scatter_ps): Ditto.
30481 (_mm512_i64scatter_pd): Ditto.
30482 (_mm512_mask_i64scatter_pd): Ditto.
30483 (_mm512_i32scatter_epi32): Ditto.
30484 (_mm512_mask_i32scatter_epi32): Ditto.
30485 (_mm512_i32scatter_epi64): Ditto.
30486 (_mm512_mask_i32scatter_epi64): Ditto.
30487 (_mm512_i64scatter_epi32): Ditto.
30488 (_mm512_mask_i64scatter_epi32): Ditto.
30489 (_mm512_i64scatter_epi64): Ditto.
30490 (_mm512_mask_i64scatter_epi64): Ditto.
30491 * config/i386/avx512pfintrin.h
30492 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
30493 (_mm512_mask_prefetch_i32gather_ps): Ditto.
30494 (_mm512_mask_prefetch_i64gather_pd): Ditto.
30495 (_mm512_mask_prefetch_i64gather_ps): Ditto.
30496 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
30497 (_mm512_prefetch_i32scatter_ps): Ditto.
30498 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
30499 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
30500 (_mm512_prefetch_i64scatter_pd): Ditto.
30501 (_mm512_prefetch_i64scatter_ps): Ditto.
30502 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
30503 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
30504 * config/i386/avx512vlintrin.h
30505 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
30506 (_mm_mmask_i32gather_ps): Ditto.
30507 (_mm256_mmask_i32gather_pd): Ditto.
30508 (_mm_mmask_i32gather_pd): Ditto.
30509 (_mm256_mmask_i64gather_ps): Ditto.
30510 (_mm_mmask_i64gather_ps): Ditto.
30511 (_mm256_mmask_i64gather_pd): Ditto.
30512 (_mm_mmask_i64gather_pd): Ditto.
30513 (_mm256_mmask_i32gather_epi32): Ditto.
30514 (_mm_mmask_i32gather_epi32): Ditto.
30515 (_mm256_mmask_i32gather_epi64): Ditto.
30516 (_mm_mmask_i32gather_epi64): Ditto.
30517 (_mm256_mmask_i64gather_epi32): Ditto.
30518 (_mm_mmask_i64gather_epi32): Ditto.
30519 (_mm256_mmask_i64gather_epi64): Ditto.
30520 (_mm_mmask_i64gather_epi64): Ditto.
30521 (_mm256_i32scatter_ps): Change __addr type to void*.
30522 (_mm256_mask_i32scatter_ps): Ditto.
30523 (_mm_i32scatter_ps): Ditto.
30524 (_mm_mask_i32scatter_ps): Ditto.
30525 (_mm256_i32scatter_pd): Ditto.
30526 (_mm256_mask_i32scatter_pd): Ditto.
30527 (_mm_i32scatter_pd): Ditto.
30528 (_mm_mask_i32scatter_pd): Ditto.
30529 (_mm256_i64scatter_ps): Ditto.
30530 (_mm256_mask_i64scatter_ps): Ditto.
30531 (_mm_i64scatter_ps): Ditto.
30532 (_mm_mask_i64scatter_ps): Ditto.
30533 (_mm256_i64scatter_pd): Ditto.
30534 (_mm256_mask_i64scatter_pd): Ditto.
30535 (_mm_i64scatter_pd): Ditto.
30536 (_mm_mask_i64scatter_pd): Ditto.
30537 (_mm256_i32scatter_epi32): Ditto.
30538 (_mm256_mask_i32scatter_epi32): Ditto.
30539 (_mm_i32scatter_epi32): Ditto.
30540 (_mm_mask_i32scatter_epi32): Ditto.
30541 (_mm256_i32scatter_epi64): Ditto.
30542 (_mm256_mask_i32scatter_epi64): Ditto.
30543 (_mm_i32scatter_epi64): Ditto.
30544 (_mm_mask_i32scatter_epi64): Ditto.
30545 (_mm256_i64scatter_epi32): Ditto.
30546 (_mm256_mask_i64scatter_epi32): Ditto.
30547 (_mm_i64scatter_epi32): Ditto.
30548 (_mm_mask_i64scatter_epi32): Ditto.
30549 (_mm256_i64scatter_epi64): Ditto.
30550 (_mm256_mask_i64scatter_epi64): Ditto.
30551 (_mm_i64scatter_epi64): Ditto.
30552 (_mm_mask_i64scatter_epi64): Ditto.
30553 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
30554 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
30555 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
30556 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
30557 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
30558 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
30559 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
30560 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
30561 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
30562 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
30563 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
30564 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
30565 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
30566 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
30567 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
30568 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
30569 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
30570 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
30571 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
30572 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
30573 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
30574 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
30575 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
30576 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
30577 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
30578 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
30579 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
30580 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
30581 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
30582 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
30583 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
30584 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
30585 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
30586 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
30587 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
30588 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
30589 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
30590 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
30591 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
30592 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
30593 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
30594 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
30595 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
30596 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
30597 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
30598 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
30599 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
30600 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
30601 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
30602 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
30603 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
30604 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
30605 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
30606 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
30607 definitions accordingly.
30609 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
30610 Kuan-Lin Chen <kuanlinchentw@gmail.com>
30613 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
30616 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
30619 * ira-conflicts.c (ira_build_conflicts): Update total conflict
30620 hard regs for inner regno.
30622 2017-01-17 Martin Liska <mliska@suse.cz>
30625 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
30626 assumption and add comment.
30628 2017-01-17 Nathan Sidwell <nathan@acm.org>
30630 * ipa-visibility.c (localize_node): New function, broken out of ...
30631 (function_and_variable_visibility): ... here. Call it.
30633 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
30635 PR middle-end/77445
30636 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
30637 correctly set frequency of oudgoing edge.
30638 (duplicate_thread_path): Fix profile updating.
30640 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30643 * configure.ac: Add GCC_BASE_VER.
30644 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
30645 version from BASE-VER file.
30646 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
30647 (gcc.o): Depend on $(BASEVER).
30648 * common.opt (dumpfullversion): New option.
30649 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
30650 * doc/invoke.texi: Document -dumpfullversion.
30651 * doc/install.texi: Document --with-gcc-major-version-only.
30652 * configure: Regenerated.
30654 2017-01-17 Richard Biener <rguenther@suse.de>
30656 PR tree-optimization/71433
30657 * tree-vrp.c (register_new_assert_for): Merge same asserts
30658 on all incoming edges.
30659 (process_assert_insertions_for): Handle insertions at the
30662 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
30664 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
30665 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
30667 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
30670 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
30673 2017-01-17 Alan Modra <amodra@gmail.com>
30676 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
30677 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
30678 symbolic stack limit when pic.
30680 2017-01-16 Martin Sebor <msebor@redhat.com>
30682 PR tree-optimization/78608
30683 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
30685 2017-01-16 Jeff Law <law@redhat.com>
30688 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
30689 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
30690 for several include directories that may be relative to sysroot.
30691 * config/i386/x-mingw32 (gplus_includedir): Define.
30692 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
30693 (native_system_includedir): Likewise.
30694 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
30695 override if TARGET_SYSTEM_ROOT is defined.
30696 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
30698 PR tree-optimization/79090
30699 PR tree-optimization/33562
30700 PR tree-optimization/61912
30701 PR tree-optimization/77485
30702 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
30703 and computed trims into the dump file.
30705 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30707 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
30709 2017-01-16 Jakub Jelinek <jakub@redhat.com>
30712 * gimplify.c (gimplify_init_constructor): If want_value and
30713 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
30717 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
30718 sequence. Formatting fixes.
30719 (doloop_optimize): Formatting fixes.
30722 * gcc.c (debug_level_greater_than_spec_func): New function.
30723 (static_spec_functions): Add debug-level-gt spec function.
30724 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
30726 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
30727 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
30728 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
30729 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
30730 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
30731 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
30733 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
30735 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
30736 QImode fixups to general and mask registers only.
30738 2017-01-16 Carl Love <cel@us.ibm.com>
30740 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
30741 for built-in functions
30742 vector signed char vec_nabs (vector signed char)
30743 vector signed short vec_nabs (vector signed short)
30744 vector signed int vec_nabs (vector signed int)
30745 vector signed long long vec_nabs (vector signed long long)
30746 vector float vec_nabs (vector float)
30747 vector double vec_nabs (vector double)
30748 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
30750 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
30751 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
30752 * doc/extend.texi: Update the documentation file for the new built-in
30755 2017-01-16 Martin Sebor <msebor@redhat.com>
30757 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
30760 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30762 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
30763 UNSPEC_VSX__XXSPLTD to require special splat handling.
30765 2017-01-16 David Malcolm <dmalcolm@redhat.com>
30768 * system.h: Poison strndup.
30770 2017-01-16 Alan Modra <amodra@gmail.com>
30773 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
30776 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
30778 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
30780 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
30782 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
30783 call recog here. Assert that INSN_CODE (insn) is non-negative.
30785 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30788 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
30790 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
30791 in the currently scheduled RTL fragment.
30793 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30795 PR rtl-optimization/78751
30796 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
30799 2017-01-14 Jeff Law <law@redhat.com>
30801 PR tree-optimization/79090
30802 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
30803 variable length stores.
30804 (compute_trims): Delete dead assignment to *trim_tail.
30805 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
30808 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
30810 PR rtl-optimization/78626
30811 PR rtl-optimization/78727
30812 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
30813 of a block, and split such blocks after everything else is finished.
30815 2017-01-14 Alan Modra <amodra@gmail.com>
30818 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
30819 target legitimate_combined_insn.
30820 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
30821 (rs6000_legitimate_combined_insn): New function.
30822 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
30824 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
30825 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
30826 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
30828 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
30830 * doc/frontends.texi (G++ and GCC): Remove references to Java.
30832 2017-01-13 Jeff Law <law@redhat.com>
30834 PR tree-optimization/33562
30835 PR tree-optimization/61912
30836 PR tree-optimization/77485
30837 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
30839 (delete_dead_assignment): Likewise.
30840 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
30841 statement to delete_dead_call and delete_dead_assignment.
30843 2017-01-13 David Malcolm <dmalcolm@redhat.com>
30846 * substring-locations.c (format_warning_va): Strengthen case 1 so
30847 that both endpoints of the substring must be within the format
30848 range for just the substring to be printed.
30850 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
30852 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
30853 * config/i386/i386.c (ix86_target_string): Add missing options
30854 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
30855 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
30856 flags_other and ix86_target_other to flags2_other. Display unknown
30858 (ix86_valid_target_attribute_inner_p): Add missing options and
30859 reorder options by implied ISAs, as in ix86_target_string.
30861 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30863 * hash-table.h (hash_table::too_empty_p): New function.
30864 (hash_table::expand): Use it.
30865 (hash_table::traverse): Likewise.
30866 (hash_table::empty_slot): Use sizeof (value_type) instead of
30867 sizeof (PTR) to convert bytes to elements. Shrink the table
30868 if the current size is excessive for the current number of
30871 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30873 * ira-costs.c (record_reg_classes): Break from the inner loop
30874 early once alt_fail is known to be true. Update outer loop
30875 handling accordingly.
30877 2017-01-13 Jeff Law <law@redhat.com>
30879 * tree-ssa-dse.c (decrement_count): New function.
30880 (increment_start_addr, maybe_trim_memstar_call): Likewise.
30881 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
30882 when we know the partially dead statement is a mem* function.
30884 PR tree-optimization/61912
30885 PR tree-optimization/77485
30886 * tree-ssa-dse.c: Include expr.h.
30887 (maybe_trim_constructor_store): New function.
30888 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
30890 PR tree-optimization/33562
30891 PR tree-optimization/61912
30892 PR tree-optimization/77485
30893 * doc/invoke.texi: Document new dse-max-object-size param.
30894 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
30895 * tree-ssa-dse.c: Include params.h.
30896 (dse_store_status): New enum.
30897 (initialize_ao_ref_for_dse): New, partially extracted from
30899 (valid_ao_ref_for_dse, normalize_ref): New.
30900 (setup_live_bytes_from_ref, compute_trims): Likewise.
30901 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
30902 (maybe_trim_partially_dead_store): Likewise.
30903 (maybe_trim_complex_store): Likewise.
30904 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
30905 Track what bytes live from the original store. Return tri-state
30906 for dead, partially dead or live.
30907 (dse_dom_walker): Add constructor, destructor and new private members.
30908 (delete_dead_call, delete_dead_assignment): New extracted from
30910 (dse_optimize_stmt): Make a member of dse_dom_walker.
30911 Use initialize_ao_ref_for_dse.
30913 PR tree-optimization/33562
30914 PR tree-optimization/61912
30915 PR tree-optimization/77485
30916 * sbitmap.h (bitmap_count_bits): Prototype.
30917 (bitmap_clear_range, bitmap_set_range): Likewise.
30918 * sbitmap.c (bitmap_clear_range): New function.
30919 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
30921 2017-01-13 Martin Liska <mliska@suse.cz>
30924 * function.c (set_cfun): Add new argument force.
30925 * function.h (set_cfun): Likewise.
30926 * ipa-inline-transform.c (inline_call): Use the function when
30927 strict alising from is dropped for function we inline to.
30929 2017-01-13 Richard Biener <rguenther@suse.de>
30931 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
30932 for dumping GIMPLE INTEGER_CSTs.
30934 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30936 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
30937 to 201112L since C++17.
30939 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
30942 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
30943 if -fsanitize=kernel-address is present.
30945 2017-01-13 Richard Biener <rguenther@suse.de>
30947 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
30948 as _Literal ( type ) number in case usual suffixes do not
30949 preserve all information.
30951 2017-01-13 Richard Biener <rguenther@suse.de>
30953 PR tree-optimization/77283
30954 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
30955 and ssa-iterators.h.
30956 (is_feasible_trace): Implement a cost model based on joiner
30959 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
30962 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
30963 char or short to __float128/_Float128 directly.
30965 2017-01-12 Martin Sebor <msebor@redhat.com>
30967 to -Wformat-overflow.
30968 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
30969 (min_bytes_remaining): Same.
30970 (get_string_length): Same.
30971 (format_string): Same.
30972 (format_directive): Same.
30974 (pass_sprintf_length::handle_gimple_call): Same.
30976 2017-01-12 Jakub Jelinek <jakub@redhat.com>
30978 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
30979 info.nowrite calls with no lhs that can't throw. Return bool
30980 whether gsi_remove has been called or not.
30981 (pass_sprintf_length::handle_gimple_call): Return bool whether
30982 try_substitute_return_value called gsi_remove. Formatting fix.
30983 (pass_sprintf_length::execute): Don't use gsi_remove if
30984 handle_gimple_call returned true.
30987 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
30988 be removed due to side-effects, don't remove following barrier nor
30989 turn the successor edge into fallthru edge.
30991 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30994 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
30995 element-reversing loads and stores as not swappable.
30997 2017-01-12 Nathan Sidwell <nathan@acm.org>
30998 Nicolai Stange <nicstange@gmail.com>
31000 * combine.c (try_combine): Don't ignore result of overlap checking
31001 loop. Combine overlap & asm check into single loop.
31003 2017-01-12 Richard Biener <rguenther@suse.de>
31005 * tree-pretty-print.c (dump_generic_node): Provide -gimple
31006 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
31008 2017-01-12 Richard Biener <rguenther@suse.de>
31010 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
31011 and TS_TARGET_OPTION directly derive from TS_BASE.
31012 * tree-core.h (tree_optimization_option): Derive from tree_base.
31013 (tree_target_option): Likewise.
31015 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
31017 * config/i386/i386.c (memory_address_length): Increase len
31018 only when rip_relative_addr_p returns false.
31020 2017-01-11 Julia Koval <julia.koval@intel.com>
31022 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
31023 (OPTION_MASK_ISA_SGX_SET): New.
31024 (ix86_handle_option): Handle OPT_msgx.
31025 * config.gcc: Added sgxintrin.h.
31026 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
31027 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
31028 * config/i386/i386.c (ix86_target_string): Add -msgx.
31030 (ix86_option_override_internal): Handle new options.
31031 (ix86_valid_target_attribute_inner_p): Add sgx.
31032 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
31033 * config/i386/i386.opt: Add msgx.
31034 * config/i386/sgxintrin.h: New file.
31035 * config/i386/x86intrin.h: Add sgxintrin.h.
31037 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31040 * fold-const.c (maybe_nonzero_address): Return 1 for function
31042 (tree_single_nonzero_warnv_p): Don't handle function local objects
31046 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
31049 2017-01-11 David Malcolm <dmalcolm@redhat.com>
31052 * opts.c: Include "spellcheck.h"
31053 (struct string_fragment): New struct.
31054 (struct edit_distance_traits<const string_fragment &>): New
31056 (get_closest_sanitizer_option): New function.
31057 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
31059 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31061 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
31063 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
31064 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
31065 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
31066 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
31067 for initial die_offset if dwarf_split_debug_info.
31068 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
31069 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
31071 (output_skeleton_debug_sections): Formatting fix. Use
31072 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
31073 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
31075 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
31077 * config/arm/cortex-a53.md: Add bypasses for
31078 cortex_a53_r2f_cvt.
31079 (cortex_a53_r2f): Only use for transfers.
31080 (cortex_a53_f2r): Likewise.
31081 (cortex_a53_r2f_cvt): Add reservation for conversions.
31082 (cortex_a53_f2r_cvt): Likewise.
31084 2017-01-11 Tamar Christina <tamar.christina@arm.com>
31086 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
31087 to all inlined functions, change static to extern.
31089 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
31092 * config/arm/arm.c (legitimize_pic_address): Handle reference to
31094 (arm_assemble_integer): Likewise.
31096 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
31098 * config.gcc: Use new awk script to check CPU, FPU and architecture
31099 parameters for --with-... options.
31100 * config/arm/parsecpu.awk: New file
31101 * config/arm/arm-cpus.in: New file.
31102 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
31104 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
31106 * config/arm/t-arm: Update dependency rules.
31107 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
31108 of processing .def files.
31109 * config/arm/genopt.sh: Deleted.
31110 * config/arm/gentune.sh: Deleted.
31111 * config/arm/arm-cores.def: Deleted.
31112 * config/arm/arm-arches.def: Deleted.
31113 * config/arm/arm-fpus.def: Deleted.
31114 * config/arm/arm-tune.md: Regenerated.
31115 * config/arm/arm-tables.opt: Regenerated.
31116 * config/arm/arm-cpu.h: New generated file.
31117 * config/arm/arm-cpu-data.h: New generated file.
31118 * config/arm/arm-cpu-cdata.h: New generated file.
31120 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
31123 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
31125 (input_varpool_node): Unpack dynamically_initialized bit.
31127 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
31129 PR rtl-optimization/79032
31130 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
31131 the alignment of the adjusted memory reference against that of MODE,
31132 instead of the alignment of the original memory reference.
31134 2017-01-11 Martin Jambor <mjambor@suse.cz>
31136 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
31138 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
31139 decorated functions.
31141 2017-01-11 Richard Biener <rguenther@suse.de>
31143 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
31144 set range/nonnull info for PHI results. Do not set it on
31145 stmts marked for removal.
31147 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
31149 * expr.c (store_field): In the bitfield case, fetch the return value
31150 from the registers before applying a single big-endian adjustment.
31151 Always do a final load for a BLKmode value not larger than a word.
31153 2017-01-10 David Malcolm <dmalcolm@redhat.com>
31156 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31157 that we correctly handle column numbers greater than
31158 LINE_MAP_MAX_COLUMN_NUMBER.
31160 2017-01-10 Martin Sebor <msebor@redhat.com>
31162 PR middle-end/78245
31163 * gimple-ssa-sprintf.c (get_destination_size): Call
31164 {init,fini}object_sizes.
31165 * tree-object-size.c (addr_object_size): Adjust.
31166 (pass_through_call): Adjust.
31167 (pass_object_sizes::execute): Adjust.
31168 * tree-object-size.h (fini_object_sizes): Declare.
31170 2017-01-10 Martin Sebor <msebor@redhat.com>
31172 PR tree-optimization/78775
31173 * builtins.c (get_size_range): Move...
31174 * calls.c: ...to here.
31175 (alloc_max_size): Accept zero argument.
31176 (operand_signed_p): Remove.
31177 (maybe_warn_alloc_args_overflow): Call get_size_range.
31178 * calls.h (get_size_range): Declare.
31180 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
31182 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
31183 from TI's devices.csv file as of September 2016.
31184 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
31186 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
31188 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
31189 * doc/invoke.texi: Likewise.
31190 * doc/md.texi: Likewise.
31191 * doc/objc.texi: Likewise.
31193 2017-01-10 Joshua Conner <joshconner@google.com>
31195 * config/arm/fuchsia-elf.h: New file.
31196 * config/fuchsia.h: New file.
31197 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
31198 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
31200 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
31202 2016-01-10 Richard Biener <rguenther@suse.de>
31204 PR tree-optimization/79034
31205 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31206 Propagate out degenerate PHIs in the joiner.
31208 2017-01-10 Martin Liska <mliska@suse.cz>
31210 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
31211 (sort_congruence_classes_by_decl_uid): Likewise.
31212 (sort_congruence_class_groups_by_decl_uid): Likewise.
31213 (sem_item_optimizer::merge_classes): Sort class, groups in these
31214 classes and members in the groups by DECL_UID of declarations.
31215 This would make merge operations stable.
31217 2017-01-10 Martin Liska <mliska@suse.cz>
31219 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
31220 usage of m_classes_vec.
31221 (sem_item_optimizer::~sem_item_optimizer): Likewise.
31222 (sem_item_optimizer::get_group_by_hash): Likewise.
31223 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
31224 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
31225 (sem_item_optimizer::verify_classes): Likewise.
31226 (sem_item_optimizer::process_cong_reduction): Likewise.
31227 (sem_item_optimizer::dump_cong_classes): Likewise.
31228 (sem_item_optimizer::merge_classes): Likewise.
31229 * ipa-icf.h (congruence_class_hash): Rename from
31230 congruence_class_group_hash. Remove declaration of m_classes_vec.
31232 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
31234 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
31235 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
31236 * config.gcc: Add avx512vpopcntdqintrin.h.
31237 * config/i386/avx512vpopcntdqintrin.h: New.
31238 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
31239 * config/i386/i386-builtin-types.def: Add new types.
31240 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
31241 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
31242 __builtin_ia32_vpopcountq_v8di_mask): New.
31243 * config/i386/i386-c.c (ix86_target_macros_internal): Define
31244 __AVX512VPOPCNTDQ__.
31245 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
31246 (PTA_AVX512VPOPCNTDQ): Define.
31247 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
31248 TARGET_AVX512VPOPCNTDQ_P): Define.
31249 * config/i386/i386.opt: Add mavx512vpopcntdq.
31250 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
31251 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
31253 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31255 PR middle-end/77484
31256 * predict.def (PRED_CALL): Set to 67.
31258 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
31260 * expr.c (store_field): In the bitfield case, if the value comes from
31261 a function call and is of an aggregate type returned in registers, do
31262 not modify the field mode; extract the value in all cases if the mode
31263 is BLKmode and the size is not larger than a word.
31265 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
31268 * config/i386/cpuid.h: Fix undefined behavior.
31270 2017-01-04 Jeff Law <law@redhat.com>
31272 PR tree-optimization/79007
31273 PR tree-optimization/67955
31274 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
31275 conservative for pt.null when flag_non_call_exceptions is on.
31277 2017-01-09 Jakub Jelinek <jakub@redhat.com>
31279 PR translation/79019
31280 PR translation/79020
31281 * params.def (PARAM_INLINE_MIN_SPEEDUP,
31282 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
31283 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
31285 * config/avr/avr.opt (maccumulate-args): Likewise.
31286 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
31287 * common.opt (freport-bug): Likewise.
31288 * cif-code.def (CIF_FINAL_ERROR): Likewise.
31289 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
31290 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
31291 translatable string.
31292 * config/i386/i386.c (function_value_32): Likewise.
31293 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
31294 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
31296 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
31297 * common/config/msp430/msp430-common.c (msp430_handle_option):
31299 * symtab.c (symtab_node::verify_base): Likewise.
31300 * opts.c (set_debug_level): Likewise.
31301 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
31302 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
31303 missing whitespace to translatable strings.
31304 * config/avr/avr.md (bswapsi2): Fix typo in comment.
31305 * config/sh/superh.h: Likewise.
31306 * config/i386/xopintrin.h: Likewise.
31307 * config/i386/znver1.md: Likewise.
31308 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
31309 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
31310 * double-int.h (struct double_int): Likewise.
31311 * double-int.c (div_and_round_double): Likewise.
31312 * wide-int.cc: Likewise.
31313 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
31314 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
31315 * cfgcleanup.c (crossjumps_occured): Renamed to ...
31316 (crossjumps_occurred): ... this.
31317 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
31320 PR tree-optimization/78899
31321 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
31322 returning bool return struct loop *, NULL for failure and the new
31324 (versionable_outer_loop_p): Don't version outer loop if it has
31325 dont_vectorized bit set.
31326 (tree_if_conversion): When versioning outer loop, ensure
31327 tree_if_conversion is performed also on the inner loop of the
31328 non-vectorizable outer loop copy.
31329 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
31330 LOOP_VECTORIZED in inner loop of the scalar outer loop and
31331 prevent vectorization of it.
31332 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
31333 the outer loop vectorization of the non-scalar version is attempted
31334 before vectorization of the inner loop in scalar version. If
31335 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
31336 vectorization of its inner loop.
31337 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
31338 has 2 inner loops, rename also on edges from bb whose single pred
31339 is outer_loop->header. Fix typo in function comment.
31341 2017-01-09 Martin Sebor <msebor@redhat.com>
31344 * asan.c (asan_emit_stack_protection): Increase local buffer size
31345 to avoid snprintf truncation warning.
31347 2017-01-09 Andrew Pinski <apinski@cavium.com>
31349 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
31350 to reference thunderx2t99 for the tuning structure
31351 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
31353 (thunderx2t99_extra_costs): This.
31354 * config/aarch64/aarch64-tune.md: Regenerate.
31355 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
31356 (vulcan_addrcost_table): This.
31357 (vulcan_regmove_cost): Rename to ...
31358 (thunderx2t99_regmove_cost): This.
31359 (vulcan_vector_cost): Rename to ...
31360 (thunderx2t99_vector_cost): this.
31361 (vulcan_branch_cost): Rename to ...
31362 (thunderx2t99_branch_cost): This.
31363 (vulcan_tunings): Rename to ...
31364 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
31365 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
31367 2017-01-09 Martin Jambor <mjambor@suse.cz>
31371 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
31372 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
31373 (propagate_vr_accross_jump_function): Use the above function for all
31374 value range computations for pass-through jump functions and type
31375 converasion from explicit value range values.
31376 (ipcp_propagate_stage): Do not attempt to deduce types of formal
31377 parameters from TYPE_ARG_TYPES.
31378 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
31379 (ipa_write_node_info): Stream type of the actual argument.
31380 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
31382 2017-01-09 Martin Liska <mliska@suse.cz>
31385 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
31386 (lookup_compiler): Do not show error message with have_E.
31388 2017-01-09 Jakub Jelinek <jakub@redhat.com>
31390 PR tree-optimization/78938
31391 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
31392 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
31393 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
31394 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
31397 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31399 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
31402 2017-01-09 Richard Biener <rguenther@suse.de>
31404 PR tree-optimization/78997
31405 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
31406 name condition properly.
31408 2017-01-09 Richard Biener <rguenther@suse.de>
31411 * dwarf2out.c (is_cxx): New overload with context.
31412 (is_naming_typedef_decl): Use it.
31414 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31416 * invoke.texi (Option Summary): Correct spacing in option lists
31417 and add line breaks to fix over-long lines.
31419 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31421 PR middle-end/17660
31423 * extend.texi (Common Variable Attributes): Add xref to GCC
31424 Internals manual to explain mode attribute keywords.
31426 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31429 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
31430 and Preprocessor Options.
31431 (Options for Linking): Document -pthread here....
31432 (RS/6000 and PowerPC Options): ...not here.
31433 (Solaris 2 Options): ...or here.
31434 * doc/cppopts.texi: Document -pthread.
31436 2017-01-08 Martin Sebor <msebor@redhat.com>
31438 PR middle-end/77708
31439 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
31440 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
31441 New member functions.
31442 (format_directive): Used them.
31444 (pass_sprintf_length::handle_gimple_call): Same.
31445 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
31446 to avoid truncation for any argument.
31447 (extract_affine_mul): Same.
31448 * tree.c (get_file_function_name): Same.
31450 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31452 PR middle-end/77484
31453 * predict.def (PRED_INDIR_CALL): Set to 86.
31455 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31457 PR preprocessor/54124
31458 * doc/cppopts.texi: Reformat -d subtable to list the full name
31459 of the options. Add cross-reference to the docs for the general
31460 compiler -d options.
31461 * doc/invoke.texi (Developer Options): Add cross-reference to the
31462 preprocessor-specific -d option documentation.
31464 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31466 PR preprocessor/13498
31467 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
31468 redudant material, and reflect new command-line options.
31469 (System Headers): Likewise.
31471 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31473 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
31474 -isystem, and -idirafter. Copy-edit.
31475 * doc/cppopts.texi: Copy-edit. Remove contradiction about
31476 default for -ftrack-macro-expansion. Delete obsolete and
31477 badly-formatted implementation details about -fdebug-cpp output.
31478 * doc/cppwarnopts.texi: Copy-edit.
31480 2017-01-07 David Malcolm <dmalcolm@redhat.com>
31483 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31484 that the transition from a max line width >= 1<<10 to narrower
31485 lines works correctly.
31487 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
31489 * doc/options.texi (PerFunction): New.
31490 * opt-functions.awk (switch_flags): Map both Optimization and
31491 PerFunction to CL_OPTIMIZATION.
31492 * opth-gen.awk: Test for PerFunction flag along with
31494 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
31495 it only when the latter is present. Skip those that don't in
31496 the hash function generator.
31497 * common.opt (fvar-tracking): Mark as PerFunction instead of
31499 (fvar-tracking-assignments): Likewise.
31500 (fvar-tracking-assignments-toggle): Likewise.
31501 (fvar-tracking-uninit): Likewise.
31503 2017-01-07 Jakub Jelinek <jakub@redhat.com>
31505 PR translation/79018
31506 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
31509 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
31512 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
31513 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
31514 TARGET_LONG_JUMP_TABLE_OFFSETS.
31515 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
31516 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
31517 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
31518 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
31519 * config/m68k/m68k.md (tablejump expander): Likewise.
31520 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
31521 TARGET_LONG_JUMP_TABLE_OFFSETS.
31522 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
31523 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
31525 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
31526 David Holsgrove <david.holsgrove@xilinx.com>
31528 * common/config/microblaze/microblaze-common.c
31529 (TARGET_EXCEPT_UNWIND_INFO): Remove.
31530 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
31532 * config/microblaze/microblaze.c (microblaze_must_save_register)
31533 (microblaze_expand_epilogue, microblaze_return_addr): Handle
31535 (microblaze_eh_return): New function.
31536 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
31537 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
31538 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
31539 * config/microblaze/microblaze.md (eh_return): New pattern.
31541 2017-01-06 Jakub Jelinek <jakub@redhat.com>
31543 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
31544 GCC_DIAGNOSTIC_STRINGIFY): Define.
31546 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
31548 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31550 * config/arm/arm.md (<mcrr>): New.
31552 * config/arm/arm.c (arm_arch5te): New.
31553 (arm_option_override): Set arm_arch5te.
31554 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
31556 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
31557 (arm_mcrr_qualifiers): ... this. New.
31558 (MRRC_QUALIFIERS): Define to...
31559 (arm_mrrc_qualifiers): ... this. New.
31560 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
31562 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
31563 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
31564 (MRRCI, mrrc, MRRC): New.
31565 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
31566 VUNSPEC_MRRC2): New.
31568 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31570 * config/arm/arm.md (<mcr>): New.
31572 * config/arm/arm.c (arm_coproc_builtin_available): Add
31573 support for mcr, mrc, mcr2 and mrc2.
31574 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
31575 (arm_mcr_qualifiers): ... this. New.
31576 (MRC_QUALIFIERS): Define to ...
31577 (arm_mrc_qualifiers): ... this. New.
31578 (MCR_QUALIFIERS): Define to ...
31579 (arm_mcr_qualifiers): ... this. New.
31580 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
31582 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
31583 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
31584 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
31585 VUNSPEC_MRC2): New.
31587 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31589 * config/arm/arm.md (*ldc): New.
31593 * config/arm/arm.c (arm_coproc_builtin_available): Add
31594 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
31595 (arm_coproc_ldc_stc_legitimate_address): New.
31596 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
31597 'qualifier_const_pointer'.
31598 (LDC_QUALIFIERS): Define to...
31599 (arm_ldc_qualifiers): ... this. New.
31600 (STC_QUALIFIERS): Define to...
31601 (arm_stc_qualifiers): ... this. New.
31602 * config/arm/arm-protos.h
31603 (arm_coproc_ldc_stc_legitimate_address): New.
31604 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
31605 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
31606 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
31607 stc2, stcl, stc2l): New.
31608 * config/arm/constraints.md (Uz): New.
31609 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
31610 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
31611 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
31612 VUNSPEC_STC2L): New.
31614 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31616 * config/arm/arm.md (<cdp>): New.
31617 * config/arm/arm.c (neon_const_bounds): Rename this ...
31618 (arm_const_bounds): ... this.
31619 (arm_coproc_builtin_available): New.
31620 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
31621 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
31622 (CDP_QUALIFIERS): Define to...
31623 (arm_cdp_qualifiers): ... this. New.
31625 (arm_expand_builtin_args): Add case for 6 arguments.
31626 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
31627 (arm_const_bounds): ... this.
31628 (arm_coproc_builtin_available): New.
31629 * config/arm/arm_acle.h (__arm_cdp): New.
31631 * config/arm/arm_acle_builtins.def (cdp): New.
31633 * config/arm/iterators.md (CDPI,CDP,cdp): New.
31634 * config/arm/neon.md: Rename all 'neon_const_bounds' to
31635 'arm_const_bounds'.
31636 * config/arm/types.md (coproc): New.
31637 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
31638 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
31639 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
31640 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
31642 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31644 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
31645 (UBINOP_QUALIFIERS): New.
31647 (acle_builtin_data): New. Change comment.
31648 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
31649 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
31650 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
31651 arm_acle_builtins.def.
31652 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
31653 (arm_init_acle_builtins): New.
31654 (CRC32_BUILTIN): Remove.
31655 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
31656 crc32cb, crc32ch and crc32cw.
31657 (arm_init_crc32_builtins): Remove.
31658 (arm_init_builtins): Use arm_init_acle_builtins rather
31659 than arm_init_crc32_builtins.
31660 (arm_expand_acle_builtin): New.
31661 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
31662 * config/arm/arm_acle_builtins.def: New.
31664 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31666 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
31667 (arm_builtin_datum): ... this.
31668 (arm_init_neon_builtin): Rename to ...
31669 (arm_init_builtin): ... this. Add a new parameters PREFIX
31670 and USE_SIG_IN_NAME.
31671 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
31672 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
31673 'arm_builtin_datum'.
31674 (arm_init_vfp_builtins): Likewise.
31675 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
31676 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
31677 (arm_expand_neon_args): Rename to ...
31678 (arm_expand_builtin_args): ... this. Rename builtin_arg
31679 enum values and differentiate between ARG_BUILTIN_MEMORY
31680 and ARG_BUILTIN_NEON_MEMORY.
31681 (arm_expand_neon_builtin_1): Rename to ...
31682 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
31683 values, arm_expand_builtin_args and add bool parameter NEON.
31684 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
31685 (arm_expand_vfp_builtin): Likewise.
31686 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
31688 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31690 PR middle-end/77484
31691 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
31692 * predict.c (tree_estimate_probability_bb): Reverse direction of
31693 polymorphic call predictor.
31695 2017-01-06 David Malcolm <dmalcolm@redhat.com>
31697 * passes.c (execute_one_pass): Split out pass-skipping logic into...
31698 (determine_pass_name_match): ...this new function and...
31699 (should_skip_pass_p): ...this new function.
31701 2017-01-06 Nathan Sidwell <nathan@acm.org>
31703 * ipa-visibility.c (function_and_variable_visibility): Reformat
31704 comments and long lines. Remove extrneous if.
31705 * symtab.c (symtab_node::make_decl_local): Fix code format.
31706 (symtab_node::set_section_for_node): Fix comment typo.
31708 2017-01-06 Martin Liska <mliska@suse.cz>
31711 * lra-constraints.c: Rename invariant to lra_invariant.
31712 * predict.c (set_even_probabilities): Initialize e to NULL.
31714 2017-01-05 Martin Sebor <msebor@redhat.com>
31716 PR tree-optimization/78910
31717 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
31718 (format_integer): Correct off-by-one error in the handling
31719 of precision with negative numbers in signed conversions..
31721 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
31723 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
31725 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31727 PR tree-optimization/71016
31728 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
31729 factor_out_conditional_conversion. Formatting fix.
31730 (factor_out_conditional_conversion): Add cond_stmt argument.
31731 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
31732 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
31735 2017-01-05 David Malcolm <dmalcolm@redhat.com>
31737 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
31738 read-rtl-function.o, and selftest-rtl.o.
31739 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
31740 (selftest::aarch64_test_loading_full_dump): New function.
31741 (selftest::aarch64_run_selftests): New function.
31742 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
31743 selftest::aarch64_run_selftests.
31744 * config/i386/i386.c
31745 (selftest::ix86_test_loading_dump_fragment_1): New function.
31746 (selftest::ix86_test_loading_call_insn): New function.
31747 (selftest::ix86_test_loading_full_dump): New function.
31748 (selftest::ix86_test_loading_unspec): New function.
31749 (selftest::ix86_run_selftests): Call the new functions.
31750 * emit-rtl.c (maybe_set_max_label_num): New function.
31751 * emit-rtl.h (maybe_set_max_label_num): New decl.
31752 * function.c (instantiate_decls): Guard call to
31753 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
31754 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
31756 * gensupport.c (gen_reader::gen_reader): Pass "false"
31757 for new "compact" param of rtx_reader.
31758 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
31759 rather than an empty string for NULL strings.
31760 * read-md.c: Potentially include config.h rather than bconfig.h.
31761 Wrap include of errors.h with #ifdef GENERATOR_FILE.
31762 (have_error): New global, copied from errors.c.
31763 (md_reader::read_name): Rename to...
31764 (md_reader::read_name_1): ...this, adding "out_loc" param,
31765 and converting "missing name or number" to returning false, rather
31767 (md_reader::read_name): Reimplement in terms of read_name_1.
31768 (md_reader::read_name_or_nil): New function.
31769 (md_reader::read_string): Handle "(nil)" by returning NULL.
31770 (md_reader::md_reader): Add new param "compact".
31771 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
31772 (md_reader::read_file): New method.
31773 * read-md.h (md_reader::md_reader): Add new param "compact".
31774 (md_reader::read_file): New method.
31775 (md_reader::is_compact): New accessor.
31776 (md_reader::read_name): Convert return type from void to file_location.
31777 (md_reader::read_name_or_nil): New decl.
31778 (md_reader::read_name_1): New decl.
31779 (md_reader::m_compact): New field.
31780 (noop_reader::noop_reader): Pass "false" for new "compact" param
31782 (rtx_reader::rtx_reader): Add new "compact" param.
31783 (rtx_reader::read_rtx_operand): Make virtual and convert return
31784 type from void to rtx.
31785 (rtx_reader::read_until): New decl.
31786 (rtx_reader::handle_any_trailing_information): New virtual function.
31787 (rtx_reader::postprocess): New virtual function.
31788 (rtx_reader::finalize_string): New virtual function.
31789 (rtx_reader::m_in_call_function_usage): New field.
31790 (rtx_reader::m_reuse_rtx_by_id): New field.
31791 * read-rtl-function.c: New file.
31792 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
31793 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
31794 (selftest::verify_three_block_rtl_cfg): New decl.
31795 * read-rtl-function.h: New file.
31796 * read-rtl.c: Potentially include config.h rather than bconfig.h.
31797 For host, include function.h, memmodel.h, and emit-rtl.h.
31798 (one_time_initialization): New function.
31799 (struct compact_insn_name): New struct.
31800 (compact_insn_names): New array.
31801 (find_code): Handle insn codes in compact dumps.
31802 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
31803 (bind_subst_iter_and_attr): Likewise.
31804 (add_condition_to_string): Likewise.
31805 (add_condition_to_rtx): Likewise.
31806 (apply_attribute_uses): Likewise.
31807 (add_current_iterators): Likewise.
31808 (apply_iterators): Likewise.
31809 (initialize_iterators): Guard usage of apply_subst_iterator with
31810 #ifdef GENERATOR_FILE.
31811 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
31812 (md_reader::read_mapping): Likewise.
31813 (add_define_attr_for_define_subst): Likewise.
31814 (add_define_subst_attr): Likewise.
31815 (read_subst_mapping): Likewise.
31816 (check_code_iterator): Likewise.
31817 (rtx_reader::read_rtx): Likewise. Move one-time initialization
31819 (one_time_initialization): New function.
31820 (rtx_reader::read_until): New method.
31821 (read_flags): New function.
31822 (parse_reg_note_name): New function.
31823 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
31824 Handle reuse_rtx ids.
31825 Wrap iterator lookup within #ifdef GENERATOR_FILE.
31826 Add parsing support for RTL dumps, mirroring the special-cases in
31827 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
31828 values, and calling handle_any_trailing_information.
31829 (rtx_reader::read_rtx_operand): Convert return type from void
31830 to rtx, returning return_rtx. Handle case 'e'. Call
31831 finalize_string on XSTR and XTMPL fields.
31832 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
31833 "(nil)" values were omitted. Call the postprocess vfunc on the
31835 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
31836 class ctor. Initialize m_in_call_function_usage. Call
31837 one_time_initialization.
31838 * rtl-tests.c (selftest::test_uncond_jump): Call
31839 set_new_first_and_last_insn.
31840 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
31841 * selftest-rtl.c: New file.
31842 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
31843 (selftest::get_insn_by_uid): New decl.
31844 * selftest-run-tests.c (selftest::run_tests): Call
31845 read_rtl_function_c_tests.
31846 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
31847 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
31850 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
31852 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
31853 operands in a special way. Assert that pos+len <= mode precision.
31855 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31857 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
31858 3 argument Alias with unlimited for the negative form.
31859 (fno-vect-cost-model): Removed.
31861 2017-01-05 Martin Liska <mliska@suse.cz>
31863 * hsa-gen.c (gen_hsa_divmod): New function.
31864 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
31866 2017-01-05 Martin Liska <mliska@suse.cz>
31869 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
31872 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31874 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
31875 small constant length operands.
31877 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31879 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
31880 between loop iterations.
31882 2017-01-05 Martin Liska <mliska@suse.cz>
31885 * gimplify.c (gimplify_decl_expr): Compare to
31886 asan_poisoned_variables instread of checking flags.
31887 (gimplify_target_expr): Likewise.
31888 (gimplify_expr): Likewise.
31889 (gimplify_function_tree): Conditionally initialize
31890 asan_poisoned_variables.
31892 2017-01-04 Jeff Law <law@redhat.com>
31894 PR tree-optimizatin/78812
31895 * rtl.h (contains_mem_rtx_p): Prototype.
31896 * ifcvt.c (containts_mem_rtx_p): Move from here to...
31897 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
31898 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
31899 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
31900 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
31902 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31904 * input.c (assert_char_at_range): Default-initialize actual_range.
31906 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31908 * df-scan.c (df_ref_create_structure): Make regno unsigned,
31909 to match the caller.
31911 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31913 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
31914 insns after final jump in test to emit dummy move.
31916 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31918 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
31919 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
31921 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31923 * multiple_target.c (create_dispatcher_calls): Init e_next.
31924 * tree-ssa-loop-split.c (split_loop): Init border.
31925 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
31928 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
31933 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
31934 (altivec_register_operand): Do not return true if the operand
31935 contains a SUBREG mixing SImode and SFmode.
31936 (vsx_register_operand): Likewise.
31937 (vsx_reg_sfsubreg_ok): New predicate.
31938 (vfloat_operand): Do not return true if the operand contains a
31939 SUBREG mixing SImode and SFmode.
31940 (vint_operand): Likewise.
31941 (vlogical_operand): Likewise.
31942 (gpc_reg_operand): Likewise.
31943 (int_reg_operand): Likewise.
31944 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
31945 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
31946 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
31948 (rs6000_emit_move_si_sf_subreg): New helper function.
31949 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
31950 fixup SUBREGs involving SImode and SFmode.
31951 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
31952 numbers for the new peephole2 optimization.
31953 (peephole2 for SFmode unions): New peephole2 to optimize cases in
31954 the GLIBC math library that do AND/IOR/XOR operations on single
31955 precision floating point.
31956 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
31957 target macros to say whether we need to avoid SUBREGs mixing
31959 (TARGET_ALLOW_SF_SUBREG): Likewise.
31960 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
31961 (UNSPEC_SI_FROM_SF): Likewise.
31962 (iorxor): Change spacing.
31963 (and_ior_xor): New iterator for AND, IOR, and XOR.
31964 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
31965 (movdi_from_sf_zero_ext): Likewise.
31966 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
31967 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
31968 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
31969 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
31970 (fms<mode>4): Likewise.
31971 (fnma<mode>4): Likewise.
31972 (fnms<mode>4): Likewise.
31973 (nfma<mode>4): Likewise.
31974 (nfms<mode>4): Likewise.
31976 2017-01-04 Marek Polacek <polacek@redhat.com>
31979 * doc/invoke.texi: Document -Wpointer-compare.
31981 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31983 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
31986 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
31987 descriptions for -gdwarf-5 and emit them as uleb128 instead of
31990 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
31993 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
31994 documentation of the powerpc_popcntb_ok attribute.
31995 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31996 code to issue warning messages if a requested CPU configuration is
31997 not supported by the binary (assembler and loader) toolchain.
31998 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
31999 made to define a built-in function that has been disabled.
32000 (paired_init_builtins): Add assertion to prevent ICE if attempt is
32001 made to define a built-in function that has been disabled.
32002 (altivec_init_builtins): Add comment explaining why definition
32003 of the DST built-in functions is not preceded by an assertion
32004 check. Add assertions to prevent ICE if attempts are made to
32005 define an altivec predicate or an abs* built-in function that has
32007 (htm_init_builtins): Add comment explaining why definition of the
32008 htm built-in functions is not preceded by an assertion check.
32010 2017-01-04 Jeff Law <law@redhat.com>
32012 PR tree-optimizatin/67955
32013 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
32014 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
32015 the points-to solution does not include pt_null. Use DECL_PT_UID
32018 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
32020 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
32021 Use gen_int_mode instead of gen_lopwart for const_int operands.
32023 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32025 PR tree-optimization/71563
32026 * match.pd: Simplify X << Y into X if Y is known to be 0 or
32027 out of range value - has low bits known to be zero.
32029 2017-01-04 Alan Modra <amodra@gmail.com>
32031 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
32032 * configure: Regenerate.
32033 * config.in: Regenerate.
32035 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32038 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
32039 a substring of the message, but strcmp with the whole message. Ifdef
32040 ENABLE_NLS, translate the message first using dgettext.
32042 2017-01-03 Jeff Law <law@redhat.com>
32044 PR tree-optimizatin/78856
32045 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
32046 (mark_threaded_blocks): Remove code to truncate thread paths that
32047 cross multiple loop headers. Instead invalidate the cached loop
32048 iteration information and handle case of a thread path walking
32049 into an irreducible region.
32051 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
32054 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
32055 assertions. Add support for doing the signbit if the IEEE 128-bit
32056 floating point value is in a GPR.
32057 * config/rs6000/rs6000.md (Fsignbit): Delete.
32058 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
32059 Update the length attribute if the value is in a GPR.
32060 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
32061 the sign or zero extension instruction, since the value is always 0/1.
32062 (signbit<mode>2_dm2): Delete using <Fsignbit>.
32065 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
32066 extracting SImode to a GPR register so that we can generate a
32067 store, limit the vector to be in a traditional Altivec register
32068 for the vextuwrx instruction.
32070 2017-01-03 Ian Lance Taylor <iant@google.com>
32072 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
32074 2017-01-03 Martin Sebor <msebor@redhat.com>
32076 PR tree-optimization/78696
32077 * gimple-ssa-sprintf.c (format_floating): Correct handling of
32078 precision. Use MPFR for %f for greater fidelity. Correct handling
32080 (pass_sprintf_length::compute_format_length): Set width and precision
32081 specified by asrerisk to void_node for vararg functions.
32082 (try_substitute_return_value): Adjust dump output.
32084 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
32086 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
32088 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
32090 * doc/invoke.texi (SPARC options): Document -mlra as the default.
32091 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
32092 -mlra/-mno-lra was passed to the compiler.
32094 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
32096 PR rtl-optimization/65618
32097 * emit-rtl.c (try_split): Move initialization of "before" and
32098 "after" to just before the call to emit_insn_after_setloc.
32100 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
32102 * doc/md.texi (Standard Names): Remove reference to Java frontend.
32104 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
32106 * dwarf2out.c (gen_enumeration_type_die): When
32107 -gno-strict-dwarf, add a DW_AT_encoding attribute.
32109 2017-01-03 Jakub Jelinek <jakub@redhat.com>
32111 PR tree-optimization/78965
32112 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
32113 Change first argument from const call_info & to call_info &. For %n
32114 set info.nowrite to false.
32116 PR middle-end/78901
32117 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
32118 possibly throwing calls.
32120 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
32121 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
32122 and fns handling, rather than in a separate case SSA_NAME.
32124 2017-01-02 Jeff Law <law@redhat.com>
32126 * config/darwin-driver.c (darwin_driver_init): Const-correctness
32127 fixes for first_period and second_period variables.
32129 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
32132 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
32133 (*insvqi_1): New insn pattern.
32134 (*insvqi_1_mem_rex64): Ditto.
32135 (*insvqi_2): Ditto.
32136 (*insvqi_3): Rename from *insvqi.
32138 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
32140 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
32142 * doc/cfg.texi (Edges): Remove reference to Java.
32143 (Maintaining the CFG): Ditto.
32145 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32147 PR middle-end/77674
32148 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
32149 transparent aliases.
32151 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32153 PR middle-end/77484
32154 * predict.def (PRED_CALL): Update hitrate.
32155 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
32156 * predict.c (tree_estimate_probability_bb): Split CALL predictor
32157 into direct/indirect/polymorphic variants.
32159 2017-01-01 Jakub Jelinek <jakub@redhat.com>
32161 Update copyright years.
32163 * gcc.c (process_command): Update copyright notice dates.
32164 * gcov-dump.c (print_version): Ditto.
32165 * gcov.c (print_version): Ditto.
32166 * gcov-tool.c (print_version): Ditto.
32167 * gengtype.c (create_file): Ditto.
32168 * doc/cpp.texi: Bump @copying's copyright year.
32169 * doc/cppinternals.texi: Ditto.
32170 * doc/gcc.texi: Ditto.
32171 * doc/gccint.texi: Ditto.
32172 * doc/gcov.texi: Ditto.
32173 * doc/install.texi: Ditto.
32174 * doc/invoke.texi: Ditto.
32176 Copyright (C) 2017 Free Software Foundation, Inc.
32178 Copying and distribution of this file, with or without modification,
32179 are permitted in any medium without royalty provided the copyright
32180 notice and this notice are preserved.