1 2018-12-10 Segher Boessenkool <segher@kernel.crashing.org>
3 * config.gcc (Obsolete configurations): Delete powerpc*-*-*spe*.
4 (Unsupported targets): Add powerpc*-*-*spe*.
5 (powerpc*-*-*spe*): Delete.
6 (powerpc-*-eabispe*): Delete.
7 (powerpc-*-rtems*spe*): Delete.
8 (powerpc*-*-linux*spe*): Delete.
9 (powerpc*-*-linux*): Do not handle the linux*spe* targets.
10 (powerpc-wrs-vxworks*spe): Delete.
11 (with_cpu setting code): Delete powerpc*-*-*spe* handling.
12 * config.host (target powerpc*-*-*spe*): Delete.
13 * doc/invoke.texi (PowerPC SPE Options): Delete.
14 (PowerPC SPE Options): Delete.
15 * config/powerpcspe: Delete.
17 2018-12-10 Uros Bizjak <ubizjak@gmail.com>
20 * config/i386/i386.c (ix86_expand_sse_cmp): For vector modes,
21 check operand 1 with vector_operand predicate.
22 (ix86_expand_sse_movcc): For vector modes, check op_true with
23 vector_operand, not nonimmediate_operand.
25 2018-12-10 Richard Biener <rguenther@suse.de>
27 * tree-affine.c (tree_to_aff_combination): Remove unreachable
29 (aff_combination_expand): Cache on SSA names, not possibly
30 on conversion trees. Avoid expanding cases we do not handle.
32 2018-12-10 Richard Biener <rguenther@suse.de>
34 PR tree-optimization/88427
35 * vr-values.c (vr_values::extract_range_from_phi_node):
36 Handle symbolic ranges conservatively when trying to drop
39 2018-12-10 Martin Jambor <mjambor@suse.cz>
42 * ipa-prop.c (determine_locally_known_aggregate_parts): Make sure
43 we check pointers against pointers.
45 2018-12-10 Richard Biener <rguenther@suse.de>
48 * gimple.c (gimple_assign_set_rhs_with_ops): Transfer EH
49 info to a newly allocated stmt.
51 2018-12-10 Jerome Lambourg <lambourg@adacore.com>
53 * config/vxworksae.h (TARGET_VXWORKS_HAVE_CTORS_DTORS): Define.
55 2018-12-10 Fredrik Nyström <fredrik@lysator.liu.se>
58 * config/sol2.h: Only use libgcc-unwind.map if
61 2018-12-10 Bin Cheng <bin.cheng@linux.alibaba.com>
63 * auto-profile.c (afdo_calculate_branch_prob): Convert profile_count
64 to afdo precision quality.
65 (afdo_callsite_hot_enough_for_early_inline): Likewise.
66 (afdo_annotate_cfg): Set profile_count for exit basic block.
68 2018-12-10 Bin Cheng <bin.cheng@linux.alibaba.com>
70 * auto-profile.c (string_table::get_index_by_decl)
71 (function_instance::get_function_instance_by_decl): Avoid infinite
72 recursion by using DECL_FROM_INLINE.
74 2018-12-09 John David Anglin <danglin@gcc.gnu.org>
76 * config.gcc (hppa*-*-linux*): Add pa/t-pa to tmake_file. Define
78 * config/pa/pa-protos.h (pa_d_target_versions): New prototype.
79 * config/pa/pa.h (TARGET_D_CPU_VERSIONS): Define.
80 * config/pa/pa-d.c: New file.
81 * config/pa/t-pa: New file.
83 2018-12-08 Bin Cheng <bin.cheng@linux.alibaba.com>
85 * ipa-cp.c (update_profiling_info): Call adjust_for_ipa_scaling for
88 2018-12-08 Jakub Jelinek <jakub@redhat.com>
91 * tree-nested.c (convert_local_reference_stmt): Handle clobbers where
92 lhs is not a decl normally, don't call use_pointer_in_frame on that
95 2018-12-08 Eric Botcazou <ebotcazou@adacore.com>
97 PR rtl-optimization/88390
98 * dwarf2cfi.c (struct dw_cfi_row): Add window_save field.
99 (cfi_row_equal_p): Compare it.
100 (dwarf2out_frame_debug_cfa_window_save): Add FAKE parameter.
101 If FAKE is false, set window_save of the current row.
102 (dwarf2out_frame_debug) <REG_CFA_TOGGLE_RA_MANGLE>: Call above
103 function with FAKE parameter set to true.
104 <REG_CFA_WINDOW_SAVE>: Likewise but with FAKE parameter set to false.
105 (change_cfi_row): Emit a DW_CFA_GNU_window_save if necessary.
107 2018-12-07 Peter Bergner <bergner@linux.ibm.com>
110 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
111 -mabi=ieeelongdouble and -mabi=ibmlongdouble without -mlong-double-128.
112 Do not error for -mabi=ibmlongdouble and no ISA 2.06 support.
113 * doc/invoke.texi: Document -mabi=ibmlongdouble and -mabi=ieeelongdouble
114 require -mlong-double-128.
116 2018-12-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
118 * config/aarch64/aarch64.c (aarch64_tuning_override_functions): Add
120 (aarch64_parse_sve_width_string): Define.
122 2018-12-07 Jeff Law <law@redhat.com>
125 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Add
126 m_update_global_ranges member. Add corresponding argument to ctor.
127 * gimple-ssa-evrp-analyze.c
128 (evrp_range_analyzer::evrp_range_analyzer): Add new argument and
129 initialize m_update_global_ranges.
130 (evrp_range_analyzer::set_ssa_range_info): Assert that we are
131 updating global ranges.
132 (evrp_range_analyzer::record_ranges_from_incoming_edge): Only
133 update global ranges if explicitly requested.
134 (evrp_range_analyzer::record_ranges_from_phis): Similarly.
135 (evrp_range_analyzer::record_ranges_from_stmt): Similarly.
136 * gimple-ssa-evrp.c (evrp_dom_walker): Pass new argument to
137 evrp_range_analyzer ctor.
138 * gimple-ssa-sprintf.c (sprintf_dom_walker): Similarly.
139 * tree-ssa-dom.c (dom_opt_dom_walker): Similarly.
141 2018-12-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
143 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum):
144 Add SVE_NOT_IMPLEMENTED value.
145 * config/aarch64/aarch64-protos.h (struct tune_params): Add sve_width
147 * config/aarch64/aarch64.c (generic_tunings,cortexa35_tunings,
148 cortexa53_tunings, cortexa57_tunings, cortexa72_tunings,
149 cortexa73_tunings, exynosm1_tunings, thunderx_tunings,
150 thunderx_tunings, tsv110_tunings, xgene1_tunings, qdf24xx_tunings,
151 saphira_tunings, thunderx2t99_tunings, emag_tunings):
153 (aarch64_estimated_poly_value): Define.
154 (TARGET_ESTIMATED_POLY_VALUE): Define.
156 2018-12-07 Paul A. Clarke <pc@us.ibm.com>
159 * config/rs6000/mmintrin.h (_mm_packs_pu16): Correctly use "__vector".
161 2018-12-07 Vladimir Makarov <vmakarov@redhat.com>
163 PR rtl-optimization/88349
164 * ira-costs.c (record_operand_costs): Check bigger reg class on
167 2018-12-07 Richard Sandiford <richard.sandiford@arm.com>
169 * config/aarch64/aarch64-sve.md (*mul<mode>3, *v<optab><mode>3):
170 Split the patterns after reload if we don't need the predicate
172 (*post_ra_mul<mode>3, *post_ra_v<optab><mode>3): New patterns.
174 2018-12-07 Richard Sandiford <richard.sandiford@arm.com>
176 * config/aarch64/iterators.md (SVE_UNPRED_FP_BINARY): New code
178 (sve_fp_op): Handle minus and mult.
179 * config/aarch64/aarch64-sve.md (*add<mode>3, *sub<mode>3)
180 (*mul<mode>3): Split the patterns after reload if we don't
181 need the predicate operand.
182 (*post_ra_<sve_fp_op><mode>3): New pattern.
184 2018-12-07 Bin Cheng <bin.cheng@linux.alibaba.com>
186 * profile-count.h (profile_count::oeprator>=): Fix typo by inverting
187 return condition when *this is precise zero.
189 2018-12-07 Jakub Jelinek <jakub@redhat.com>
192 * final.c (rest_of_handle_final): Don't call collect_fn_hard_reg_usage
193 for functions with naked attribute.
196 * tree-vrp.c (extract_range_from_binary_expr): For POINTER_PLUS_EXPR
197 with -fno-delete-null-pointer-checks, set_nonnull only if the pointer
198 is non-NULL and offset is known to have most significant bit clear.
199 * vr-values.c (vr_values::vrp_stmt_computes_nonzero): For ADDR_EXPR
200 of MEM_EXPR, return true if the MEM_EXPR has non-zero offset with
201 most significant bit clear. If offset does have most significant bit
202 set and -fno-delete-null-pointer-checks, don't return true even if
203 the base pointer is non-NULL.
205 2018-12-06 Alexandre Oliva <aoliva@redhat.com>
207 * cselib.c (cselib_record_sets): Skip strict low part sets
210 2018-12-06 Paul A. Clarke <pc@us.ibm.com>
213 * config/rs6000/smmintrin.h: New file.
214 * config.gcc: Add smmintrin.h to extra_headers for powerpc*-*-*.
216 2018-12-06 Paul A. Clarke <pc@us.ibm.com>
219 * config/rs6000/mmintrin.h (_mm_unpackhi_pi8): Fix for big-endian.
220 (_mm_unpacklo_pi8): Likewise.
221 (_mm_mulhi_pi16): Likewise.
222 (_mm_packs_pi16): Fix for big-endian. Use preferred API.
223 (_mm_packs_pi32): Likewise.
224 (_mm_packs_pu16): Likewise.
225 * config/rs6000/xmmintrin.h (_mm_cvtss_si32): Fix for big-endian.
226 (_mm_cvtss_si64): Likewise.
227 (_mm_cvtpi32x2_ps): Likewise.
228 (_mm_shuffle_ps): Likewise.
229 (_mm_movemask_pi8): Likewise.
230 (_mm_mulhi_pu16): Likewise.
231 (_mm_sad_pu8): Likewise.
232 (_mm_sad_pu8): Likewise.
233 (_mm_cvtpu16_ps): Fix for big-endian. Use preferred API.
234 (_mm_cvtpu8_ps): Likewise.
235 (_mm_movemask_ps): Better #else case for big-endian (no functional
237 (_mm_shuffle_pi16): Likewise.
238 * config/rs6000/emmintrin.h (_mm_movemask_pd): Fix for big-endian.
239 Better #else case for big-endian (no functional change).
240 (_mm_movemask_epi8): Likewise.
241 (_mm_shufflehi_epi16): Likewise.
242 (_mm_shufflelo_epi16): Likewise.
243 (_mm_shuffle_epi32): Likewise.
244 (_mm_mul_epu32): Fix for big-endian.
245 (_mm_bsrli_si128): Likewise.
246 (_mm_cvtps_pd): Better #else case for big endian.
247 (_mm_mulhi_epi16): Likewise.
248 (_mm_mul_epu32): Likewise.
249 (_mm_slli_si128): Likewise.
250 (_mm_sll_epi16): Likewise.
251 (_mm_sll_epi32): Likewise.
252 (_mm_sra_epi16): Likewise.
253 (_mm_sra_epi32): Likewise.
254 (_mm_srl_epi16): Likewise.
255 (_mm_srl_epi32): Likewise.
256 (_mm_mulhi_epu16): Likewise.
257 (_mm_sad_epu8): Likewise.
258 * config/rs6000/pmmintrin.h (_mm_hadd_ps): Fix for big-endian.
259 (_mm_sub_ps): Likewise.
260 * config/rs6000/mmintrin.h (_mm_cmpeq_pi8): Fix for 32-bit mode.
261 * gcc/config/rs6000/tmmintrin.h (_mm_alignr_epi8): Use ENDIAN
262 macros consistently (no functional changes).
263 (_mm_alignr_pi8): Likewise.
265 2018-12-06 Iain Sandoe <iain@sandoe.co.uk>
268 * config/darwin.h (TARGET_WEAK_NOT_IN_ARCHIVE_TOC) Remove, use the
270 * config/rs6000/darwin7.h (TARGET_WEAK_NOT_IN_ARCHIVE_TOC): New.
272 2018-12-06 Vladimir Makarov <vmakarov@redhat.com>
275 * ira.c (ira_init_register_move_cost): Use info from
276 hard_regno_mode_ok instead of contains_reg_of_mode.
277 * ira-costs.c (contains_reg_of_mode): Don't use cost from bigger
278 hard register class for some fixed hard registers.
280 2018-12-06 Segher Boessenkool <segher@kernel.crashing.org>
282 * doc/extend.texi (Using Assembly Language with C): Document asm inline.
283 (Size of an asm): Fix typo. Document asm inline.
284 * gimple-pretty-print.c (dump_gimple_asm): Handle asm inline.
285 * gimple.h (enum gf_mask): Add GF_ASM_INLINE.
286 (gimple_asm_set_volatile): Fix typo.
287 (gimple_asm_inline_p): New.
288 (gimple_asm_set_inline): New.
289 * gimplify.c (gimplify_asm_expr): Propagate the asm inline flag from
291 * ipa-icf-gimple.c (func_checker::compare_gimple_asm): Compare the
292 gimple_asm_inline_p flag, too.
293 * tree-core.h (tree_base): Document that protected_flag is ASM_INLINE_P
295 * tree-inline.c (estimate_num_insns): If gimple_asm_inline_p return
296 a minimum size for an asm.
297 * tree.h (ASM_INLINE_P): New.
299 2018-12-06 Segher Boessenkool <segher@kernel.crashing.org>
302 * doc/extend.texi (Basic Asm): Update grammar.
303 (Extended Asm): Update grammar.
305 2018-12-06 Iain Sandoe <iain@sandoe.co.uk>
308 * config/i386/darwin.h (STACK_BOUNDARY): Remove macro.
309 * config/i386/i386.c (ix86_compute_frame_layout): Ensure at least 128b
310 stack alignment in non-leaf functions.
312 2018-12-06 Jakub Jelinek <jakub@redhat.com>
315 * config/aarch64/aarch64.c (aarch64_print_address_internal): Don't
316 call output_operand_lossage on VOIDmode CONST_INTs. After
317 output_operand_lossage do return false.
319 2018-12-06 Richard Biener <rguenther@suse.de>
321 * df-problems.c (df_rd_local_compute): Use bitmap_release.
322 (df_live_free): Likewise.
323 (df_md_local_compute): Likewise.
324 (df_md_free): Release df_md_scratch bitmap.
325 * loop-invariant.c (calculate_loop_reg_pressure): Use
327 * sched-deps.c (true_dependency_cache, output_dependency_cache,
328 anti_dependency_cache, control_dependency_cache,
329 spec_dependency_cache): Use bitmap instead of bitmap_head *.
330 * sched-ebb.c (schedule_ebbs_init): Initialize non-GTY
331 dont_calc_deps as bitmap allocated from obstack not GC.
332 (schedule_ebbs_finish): Use bitmap_release.
333 * sched-rgn.c (schedule_insns): Initialize non-GTY
334 not_in_df as bitmap allocated from obstack not GC.
336 * sel-sched.c (_forced_ebb_heads): Remove premature optimization.
337 (sel_region_init): Allocate forced_ebb_heads.
338 (sel_region_finish): Free forced_ebb_heads.
340 2018-12-06 Richard Biener <rguenther@suse.de>
342 * bitmap.c (bitmap_head::crashme): Define.
343 * bitmap.h (bitmap_head): Add constexpr default constructor
344 poisoning the obstack member.
345 (bitmap_head::crashme): Declare.
346 (bitmap_release): New function clearing a bitmap and poisoning
348 * gengtype.c (main): Make it recognize CONSTEXPR.
349 * lra-constraints.c (lra_inheritance): Use bitmap_release
350 instead of bitmap_clear.
351 * ira.c (ira): Work around class-memaccess warning.
352 * regrename.c (create_new_chain): Likewise.
354 2018-12-06 Jakub Jelinek <jakub@redhat.com>
356 PR tree-optimization/85726
357 * generic-match-head.c (optimize_successive_divisions_p): New function.
358 * gimple-match-head.c (optimize_successive_divisions_p): Likewise.
359 * match.pd: Don't combine successive divisions if they aren't exact
360 and optimize_successive_divisions_p is false.
363 * tree-nested.c (convert_nonlocal_reference_stmt): Remove clobbers
364 for non-local automatic decls.
366 2018-12-05 David Edelsohn <dje.gcc@gmail.com>
368 * config/rs6000/aix72.h (ASM_DEFAULT_SPEC): Match Power7 processor
371 2018-12-05 Uros Bizjak <ubizjak@gmail.com>
373 * dwarf2out.c (field_byte_offset): Remove
374 #ifdef PCC_BITFIELD_TYPE_MATTERS check.
376 2018-12-05 Richard Biener <rguenther@suse.de>
377 Martin Sebor <msebor@redhat.com>
380 * calls.c (get_attr_nonstring_decl): Avoid setting *REF to
382 * gcc/gimple-low.c (lower_stmt): Fold builtin calls here.
383 * gimplify (maybe_fold_stmt): Avoid folding builtin calls.
385 2018-12-05 Iain Sandoe <iain@sandoe.co.uk>
387 * configure.ac (gcc_cv_otool): Set.
388 * configure: Regenerate.
390 2018-12-05 Uros Bizjak <ubizjak@gmail.com>
392 * config/i386/cygming.h (PCC_BITFIELD_TYPE_MATTERS): Remove.
393 (GROUP_BITFIELDS_BY_ALIGN): Ditto.
395 2018-12-05 Richard Sandiford <richard.sandiford@arm.com>
397 PR tree-optimization/88064
398 * tree-vect-patterns.c (vect_convert_input): Convert the result of
399 an existing cast if it has the right width but the wrong sign.
400 Do not test the signedness of the required result when
401 considering whether to split an existing cast; instead split to
402 a type with the same signedness as the source of the cast, then
403 convert it to the opposite signedness where necessary.
404 (vect_recog_over_widening_pattern): Handle sign changes between
405 the final PLUS_EXPR and the RSHIFT_EXPR.
406 (vect_recog_average_pattern): Use an unsigned operation when
407 truncating an addition, subtraction or multiplication. Cast the
408 result back to the "real" signedness before promoting.
410 2018-12-05 Richard Biener <rguenther@suse.de>
412 PR tree-optimization/86637
413 * tree-vectorizer.c (pass_slp_vectorize::execute): Reset
414 vect_location at the end.
416 2018-12-05 Jakub Jelinek <jakub@redhat.com>
418 PR tree-optimization/87360
419 * gimple-loop-jam.c (tree_loop_unroll_and_jam): On failure to analyze
420 data dependencies, don't return false, just continue. Formatting
422 (merge_loop_tree, bb_prevents_fusion_p, unroll_jam_possible_p,
423 fuse_loops): Formatting fixes.
426 * cfgexpand.c (expand_stack_vars): If asan_vec is empty, start with
427 aligning frame offset to ASAN_RED_ZONE_SIZE bytes.
429 2018-12-05 Xianmiao Qu <xianmiao_qu@c-sky.com>
431 * config.gcc (csky-*-linux-gnu*): Force .init_array support.
433 2018-12-05 Segher Boessenkool <segher@kernel.crashing.org>
435 * config/rs6000/rs6000.c (rs6000_function_arg): Only do the special
436 aggregate handling on actual AIX, not on somewhat similar systems.
437 (rs6000_arg_partial_bytes): Ditto.
439 2018-12-04 Vladimir Makarov <vmakarov@redhat.com>
441 PR rtl-optimization/88317
442 * lra-constraints.c (split_reg): Don't set up check_only_regs if
443 we are outside the inheritance pass.
445 2018-12-04 Jan Hubicka <hubicka@ucw.cz>
447 * ipa-prop.c (jump_function_useful_p): New.
448 (ipa_read_node_info): Do not allocated useless jump functions.
450 2018-12-04 David Malcolm <dmalcolm@redhat.com>
453 * spellcheck.c (selftest::test_suggestions): Add another assertion.
455 2018-12-04 Martin Sebor <msebor@redhat.com>
457 * doc/extend.texi (__builtin_has_attribute): Fix typo.
459 2018-12-04 Jakub Jelinek <jakub@redhat.com>
462 * config/rs6000/rs6000.c (ccr_bit): Return -1 instead of assertion
464 (print_operand): Use REG_P instead of GET_CODE == REG.
465 <case 'D'>: Also check CR_REGNO_P (REGNO (x)).
466 <case 't'>: Likewise. Remove GET_MODE check.
467 * config/rs6000/rs6000.md (scc patterns): Assert ccr_bit didn't
470 2018-12-04 Martin Sebor <msebor@redhat.com>
472 * doc/extend.texi (attribute aligned): Expand.
474 2018-12-04 David Edelsohn <dje.gcc@gmail.com>
477 * config/rs6000/rs6000.c (rs6000_function_arg): Don't pass aggregates
479 (rs6000_arg_partial_bytes): Same.
481 2018-12-04 Vladimir Makarov <vmakarov@redhat.com>
484 * ira-costs.c (exec): Try bigger class to use smaller register
487 2018-12-04 Michael Ploujnikov <michael.ploujnikov@oracle.com>
490 * ipa-cp.c (create_specialized_node): Track clone counters by
491 node assembler names.
492 (ipcp_driver): Change type of clone_num_suffixes key to const char*.
494 2018-12-04 Richard Biener <rguenther@suse.de>
496 * tree-vect-loop.c (vect_transform_loop): Fix epilogue iterations
497 for epilogue vectorization.
499 2018-12-04 Richard Biener <rguenther@suse.de>
501 PR tree-optimization/88301
502 * tree-vrp.c (register_edge_assert_for_2): Fix sign-conversion
503 issues in last commit.
505 2018-12-04 Richard Biener <rguenther@suse.de>
507 PR tree-optimization/88315
508 * tree-vect-loop.c (get_initial_defs_for_reduction): Simplify
509 and fix initialization vector for SAD and DOT_PROD SLP reductions.
511 2018-12-03 Sandra Loosemore <sandra@codesourcery.com>
514 * doc/extend.texi (Nonlocal gotos): New section.
516 2018-12-03 Uros Bizjak <ubizjak@gmail.com>
518 * config/i386/sse.md (vec_concatv2df): Change (v,xm,C)
519 alternative to (v,vm,c).
521 2018-12-03 Segher Boessenkool <segher@kernel.crashing.org>
523 * gcc/config.gcc (Unsupported targets): Add powerpc*-*-linux*paired*.
524 (powerpc*-*-linux*): Don't handle it here.
526 2018-12-03 Jeff Law <law@redhat.com>
528 * config/h8300/h8300.md (<code>qi3_1): Add gcc_unreachable to keep
531 * config/h8300/h8300.md (low byte masking peepholes): Only accept
532 nonimmediate_operand for the to-be-masked operand.
534 2018-12-03 Richard Biener <rguenther@suse.de>
536 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
537 allow expected function-pointer cast re-instantiation.
539 2018-12-03 Ilya Leoshkevich <iii@linux.ibm.com>
541 * common/config/s390/s390-common.c (s390_option_init_struct):
542 Use section anchors by default.
543 * config/s390/s390.c (s390_check_symref_alignment): Handle
545 (TARGET_MAX_ANCHOR_OFFSET): Use short displacement.
546 * output.h (assemble_align): Pass `align' as unsigned int, so
547 that the value 0x80000000, which corresponds to `aligned(1 <<
548 28)', would pass the `align > BITS_PER_UNIT' check.
549 * varasm.c (assemble_align): Likewise.
551 2018-12-03 Julian Brown <julian@codesourcery.com>
553 * tree-pretty-print.c (dump_omp_clause): Make default case
556 2018-12-03 Richard Biener <rguenther@suse.de>
558 PR tree-optimization/88301
559 * tree-vrp.c (register_edge_assert_for_2): Handle conversions
560 that do not change the value by registering the same assert
563 2018-12-03 Richard Biener <rguenther@suse.de>
565 * tree-vect-loop.c (vect_transform_loop): Properly compute
566 upper bound for the epilogue when doing epilogue vectorization.
568 2018-12-03 Ilya Leoshkevich <iii@linux.ibm.com>
571 * cfgcleanup.c (class pass_postreload_jump): New pass.
572 (pass_postreload_jump::execute): Likewise.
573 (make_pass_postreload_jump): Likewise.
574 * passes.def: Add pass_postreload_jump before
576 * tree-pass.h (make_pass_postreload_jump): New pass.
578 2018-12-02 Uros Bizjak <ubizjak@gmail.com>
580 * config/i386/i386.c (inline_memory_move_cost): Correctly
581 compute number of TFmode GPR moves.
583 2018-12-02 Jakub Jelinek <jakub@redhat.com>
586 * config/i386/sse.md (*vec_concatv4sf_0, *vec_concatv4si_0): New insns.
588 2018-12-02 Jeff Law <law@redhat.com>
590 * config/h8300/h8300.md (call, call_value): Drop mode from
591 operand containing number of bytes pushed.
592 (call_insn, call_value_insn): Use mode iterator for number
593 of bytes pushed operand. Make pattern name vary appropriately.
595 2018-12-02 Jan Hubicka <jh@suse.cz>
597 * bb-reorer.c (find_rarely_executed_basic_blocks_and_crossin_edges):
598 trust precise profiles.
600 2018-12-02 Jakub Jelinek <jakub@redhat.com>
602 * tree-nested.c (convert_nonlocal_omp_clauses,
603 convert_local_omp_clauses): Handle OMP_CLAUSE_IN_REDUCTION,
604 OMP_CLAUSE_TASK_REDUCTION and OMP_CLAUSE__SIMT_ clauses.
605 (convert_nonlocal_reference_stmt, convert_local_reference_stmt):
606 Convert clauses for GIMPLE_OMP_TASKGROUP.
608 * omp-low.c (check_omp_nesting_restrictions): Allow cancel or
609 cancellation point with taskgroup clause inside of taskloop. Consider
610 a taskloop construct without nogroup clause as implicit taskgroup for
611 diagnostics if cancel/cancellation point with taskgroup clause is
612 closely nested inside of taskgroup region.
615 * asan.c (asan_clear_shadow): Move assert that len is multiple of 4
616 to the start of the function.
617 (asan_emit_stack_protection): When emitting clearing sequence for
618 epilogue, align offset down to ASAN_RED_ZONE_SIZE granularity,
619 add last_size_aligned which is last_size padded to multiples of
620 ASAN_RED_ZONE_SIZE and emit asan_clear_shadow always on 4 byte
623 2018-12-01 Jakub Jelinek <jakub@redhat.com>
626 * asan.c (asan_redzone_buffer::flush_redzone_payload): Fix up
627 an off-by-one for BYTES_BIG_ENDIAN.
630 * combine.c (find_split_point): For invalid memory address
631 nonobj + obj + const, if reg + obj + const is valid addressing
632 mode, split at nonobj. Use if rather than else if for the
633 fallback. Comment fixes.
635 2018-11-30 Indu Bhagat <indu.bhagat@oracle.com>
637 * coverage.c (get_coverage_counts): Use from_function_decl for precise
639 * profile-count.c (profile_count::dump): Add handling for precise
641 * profile.c (compute_branch_probabilities): Rely on exec_counts instead
642 of profile_info to set profile_status of function.
643 (branch_prob): Do not set profile_status of function based on
644 profile_info. Done above based on exec_counts.
646 2018-11-30 Jakub Jelinek <jakub@redhat.com>
648 PR tree-optimization/88274
649 * tree-ssa-reassoc.c (optimize_range_tests_xor,
650 optimize_range_tests_diff): If type has smaller precision than
651 corresponding mode or if it has non-standard min/max, compute
652 everything in a standard type for the precision.
655 * params.def (PARAM_LOGICAL_OP_NON_SHORT_CIRCUIT): New param.
656 * tree-ssa-ifcombine.c (ifcombine_ifandif): If
657 --param logical-op-non-short-circuit is present, override
658 LOGICAL_OP_NON_SHORT_CIRCUIT value from the param.
659 * fold-const.c (fold_range_test, fold_truth_andor): Likewise.
661 2018-11-30 Jeff Law <law@redhat.com>
663 * optabs.c (expand_binop): Use "machine_mode" rather than
664 "enum machine mode" in most recent change.
666 2018-11-30 Wilco Dijkstra <wdijkstr@arm.com>
669 * builtins.c (expand_builtin_longjmp): Use a temporary when restoring
671 (expand_builtin_nonlocal_goto): Likewise.
673 2018-11-30 David Malcolm <dmalcolm@redhat.com>
675 * diagnostic-core.h (emit_diagnostic): New decl.
676 * diagnostic.c (emit_diagnostic): New overload, taking a
679 2018-11-30 David Malcolm <dmalcolm@redhat.com>
681 * pretty-print.c (class selftest::test_pretty_printer): New
682 subclass of pretty_printer.
683 (selftest::test_prefixes_and_wrapping): New test.
684 (selftest::pretty_print_c_tests): Call it.
686 2018-11-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
688 Minimize clone counter memory usage in create_virtual_clone.
689 * cgraph.h (clone_function_name): Add a variant that takes a
691 * cgraph.h (cgraph_node::create_virtual_clone): Add a new
692 argument: num_suffix.
693 * cgraphclones.c (cgraph_node::create_virtual_clone): Pass
694 num_suffix to clone_function_name.
695 (clone_function_name): Add a variant that takes a tree decl.
696 * ipa-cp.c (create_specialized_node): Keep track of clone
697 counters in clone_num_suffixes hash map.
698 (ipcp_driver): Free the counter hash map.
699 * ipa-hsa.c (process_hsa_functions): Creates at most one hsa
702 2018-11-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
703 Make function assembly more independent.
705 This is achieved by having clone_function_name assign unique clone
706 numbers for each function independently.
708 * cgraphclones.c: Replaced clone_fn_id_num with clone_fn_ids;
710 (clone_function_name_numbered): Use clone_fn_ids.
712 2018-11-30 Vladimir Makarov <vmakarov@redhat.com>
714 PR rtl-optimization/88179
715 * lra-constraints.c (address_eliminator::address_eliminator):
716 Don't eleminate regs in illegitimate address.
718 2018-11-30 David Malcolm <dmalcolm@redhat.com>
720 PR preprocessor/88257
721 * input.c (get_substring_ranges_for_loc): Fix indentation.
722 Bulletproof against getting a different files back from
723 linemap_resolve_location and expand_location_to_spelling_point.
725 2018-11-30 Alexander Monakov <amonakov@ispras.ru>
727 PR gcov-profile/88279
728 * tree-profile.c (init_ic_make_global_vars): Call
729 decl_default_tls_model for variable and not it's type.
731 2018-11-30 Sam Tebbs <sam.tebbs@arm.com>
733 * config/arm/arm.h (TARGET_DOTPROD): Add TARGET_VFP5 constraint.
735 2018-11-30 Martin Liska <mliska@suse.cz>
738 * asan.c (asan_shadow_cst): Remove, partially transform
739 into flush_redzone_payload.
740 (RZ_BUFFER_SIZE): New.
741 (struct asan_redzone_buffer): New.
742 (asan_redzone_buffer::emit_redzone_byte): Likewise.
743 (asan_redzone_buffer::flush_redzone_payload): Likewise.
744 (asan_redzone_buffer::flush_if_full): Likewise.
745 (asan_emit_stack_protection): Use asan_redzone_buffer class
746 that is responsible for proper aligned stores and flushing
747 of shadow memory payload.
748 * asan.h (ASAN_MIN_RED_ZONE_SIZE): New.
749 (asan_var_and_redzone_size): Likewise.
750 * cfgexpand.c (expand_stack_vars): Use smaller alignment
751 (ASAN_MIN_RED_ZONE_SIZE) in order to make shadow memory
752 for automatic variables more compact.
754 2018-11-30 Alan Modra <amodra@gmail.com>
756 * config/rs6000/predicates.md (easy_fp_constant): Avoid long
757 dependent insn sequences.
758 * config/rs6000/rs6000.c (num_insns_constant): Support long
760 * config/rs6000/rs6000.md (mov<mode>_softfloat <FMOVE128>): Adjust
763 2018-11-30 Alan Modra <amodra@gmail.com>
765 * config/rs6000/rs6000.c (num_insns_constant_gpr): Renamed from
766 num_insns_constant_wide. Make static. Revise comment.
767 (num_insns_constant_multi): New function.
768 (num_insns_constant): Formatting. Correct CONST_WIDE_INT
769 calculation. Simplify and extract code common to both
770 CONST_INT and CONST_DOUBLE. Add gcc_unreachable for unhandled
772 * config/rs6000/rs6000-protos.h (num_insns_const_wide): Delete.
774 2018-11-30 Richard Biener <rguenther@suse.de>
776 PR tree-optimization/88274
777 * tree-vrp.c (ranges_from_anti_range): Fix handling of
780 2018-11-29 Andi Kleen <ak@linux.intel.com>
782 * config/i386/i386.c (current_fentry_section): Add.
783 (x86_function_profiler): Handle fentry section.
784 (ix86_attribute_table): Add fentry section.
785 * config/i386/i386.opt: Add -mfentry-section.
786 * doc/extend.texi: Document fentry_section attribute.
787 * doc/invoke.texi: Document -mfentry-section.
789 2018-11-29 Andi Kleen <ak@linux.intel.com>
791 * config/i386/i386.c (x86_print_call_or_nop): Handle nop name.
792 (current_fentry_name): Add.
793 (x86_function_profiler): Handle fentry_name attribute.
794 (ix86_handle_fentry_name): Add.
795 (ix86_attribute_table): Add fentry_name.
796 * config/i386/i386.opt: Add -mfentry-name
797 * doc/extend.texi: Document fentry_name.
798 * doc/invoke.texi: Document minstrument-return.
800 2018-11-29 Andi Kleen <ak@linux.intel.com>
802 * config/i386/i386-opts.h (enum instrument_return): Add.
803 * config/i386/i386.c (output_return_instrumentation): Add.
804 (ix86_output_function_return): Call output_return_instrumentation.
805 (ix86_output_call_insn): Call output_return_instrumentation.
806 * config/i386/i386.opt: Add -minstrument-return=.
807 * doc/invoke.texi (-minstrument-return): Document.
809 2018-11-29 Eric Botcazou <ebotcazou@adacore.com>
812 * config/sparc/sparc-modes.def: Minor tweak.
813 * config/sparc/sparc.c: Minor reordering.
814 (sparc_pass_by_reference): Move around.
815 (traverse_record_type): Change offset from HOST_WIDE_INT to int.
816 (classify_registers): Likewise for bitpos.
817 (function_arg_slotno): Remove dead test and tweak comments.
818 <MODE_RANDOM>: Remove useless assertion and test whether the
819 parameter is named in order to pass it in FP registers. Return
820 the regno for floating-point vector types.
821 (compute_int_layout): Change bitpos from HOST_WIDE_INT to int.
822 (compute_fp_layout): Likewise.
823 (count_registers): Likewise.
824 (assign_int_registers): Likewise.
825 (assign_fp_registers): Likewise.
826 (assign_registers): Likewise.
827 (function_arg_record_value): Change size from HOST_WIDE_INT to int
828 and use CEIL_NWORDS to compute the number of registers.
829 (function_arg_union_value): Minor tweaks.
830 (function_arg_vector_value): Add slotno and named parameters, use
831 CEIL_NWORDS to compute the number of registers.
832 (sparc_function_arg_1): Rework handling of vector types. Change
833 size from HOST_WIDE_INT to int.
834 (sparc_arg_partial_bytes): Rework handling of 32-bit ABI and deal
835 with vector types for the 64-bt ABI.
836 (sparc_function_arg_advance): Likewise.
837 (sparc_return_in_memory): Add reference to -fpcc-struct-return.
838 (sparc_struct_value_rtx): Return NULL_RTX instead of 0.
839 (sparc_function_value_1): Rework handling of vector types. Change
840 size from HOST_WIDE_INT to int.
842 2018-11-29 Jakub Jelinek <jakub@redhat.com>
845 * config/i386/sse.md (*<sse>_movmsk<ssemodesuffix><avxsizesuffix>_lt,
846 *<sse>_movmsk<ssemodesuffix><avxsizesuffix>_zext_lt,
847 *<sse>_movmsk<ssemodesuffix><avxsizesuffix>_shift,
848 *<sse>_movmsk<ssemodesuffix><avxsizesuffix>_zext_shift,
849 *<sse2_avx2>_pmovmskb_lt, *<sse2_avx2>_pmovmskb_zext_lt): New
850 define_insn_and_split patterns.
854 (*<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>_lt,
855 *<sse4_1>_blendv<ssefltmodesuffix><avxsizesuffix>_ltint,
856 *<sse4_1_avx2>_pblendvb_lt): Change define_insn into
857 define_insn_and_split.
859 2018-11-29 Peter Bergner <bergner@linux.ibm.com>
862 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
863 -mabi=ieeelongdouble without both -mpopcntd and -mvsx.
865 2018-11-29 Uros Bizjak <ubizjak@gmail.com>
867 * config/i386/i386.c (inline_memory_move_cost):
868 Check "in" for 2 in MMX_CLASS_P case.
869 * config/i386/mmx.md (*mov<mode>_internal): Correct
870 TARGET_INTER_UNIT_MOVES_FROM_VEC and TARGET_INTER_UNIT_MOVES_TO_VEC
871 alternatives in preferred_for_speed attribute calculation.
873 2018-11-29 Martin Sebor <msebor@redhat.com>
877 * doc/extend.texi (attribute constructor): Clarify.
879 2018-11-29 Martin Liska <mliska@suse.cz>
882 * builtins.c (expand_movstr): Fix thinko introduced
883 when switching to the new enum.
885 2018-11-29 qing zhao <qing.zhao@oracle.com>
887 * cif-code.def (EXTERN_LIVE_ONLY_STATIC): New CIF code.
888 * common.opt: Add -flive-patching flag.
889 * doc/invoke.texi: Document -flive-patching.
890 * flag-types.h (enum live_patching_level): New enum.
891 * ipa-inline.c (can_inline_edge_p): Disable external functions from
892 inlining when flag_live_patching is LIVE_PATCHING_INLINE_ONLY_STATIC.
893 * opts.c (control_options_for_live_patching): New function.
894 (finish_options): Make flag_live_patching incompatible with flag_lto.
895 Control IPA optimizations based on different levels of
898 2018-11-29 Giuliano Belinassi <giuliano.belinassi@usp.br>
900 * match.pd (sinh (atanh (x))): New simplification rules.
901 (cosh (atanh (x))): Likewise.
903 2018-11-29 Jakub Jelinek <jakub@redhat.com>
906 * config/i386/sse.md (ssebytemode): Add V16SI, V8SI and V4SI entries.
907 (ssefltmodesuffix, ssefltvecmode): New define_mode_attrs.
908 (*<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>_lt,
909 *<sse4_1>_blendv<ssefltmodesuffix><avxsizesuffix>_ltint,
910 *<sse4_1_avx2>_pblendvb_lt): New define_insns.
913 * tree.h (build_uniform_cst, uniform_integer_cst_p): Declare.
914 * tree.c (build_uniform_cst, uniform_integer_cst_p): New functions.
915 * match.pd (define_predicates): Add uniform_integer_cst_p.
916 (cmp @0 INTEGER_CST@1, cmp (convert?@2 @0) INTEGER_CST@1): Adjust
917 so that it works also for vector comparisons with uniform constants
918 with INTEGER_CST element.
921 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): For
922 vec_add and vec_sub builtins, perform PLUS_EXPR or MINUS_EXPR
923 in unsigned_type_for instead of vector integral type where overflow
926 2018-11-29 Michael Ploujnikov <michael.ploujnikov@oracle.com>
928 There can be at most one .resolver clone per function
929 * config/rs6000/rs6000.c (make_resolver_func): Generate
930 resolver symbol with clone_function_name instead of
931 clone_function_name_numbered.
933 2018-11-29 Richard Biener <rguenther@suse.de>
935 PR tree-optimization/88243
936 * tree-vect-patterns.c (vect_mark_pattern_stmts): Set the def
937 type of all pattern-sequence stmts to vect_internal_def.
939 2018-11-29 Andre Vieira <andre.simoesdiasvieira@arm.com>
942 * config/arm/arm-cpus.in (armv7-r): Add FP16conv configurations.
943 (cortex-r7, cortex-r8): Update default and add new configuration.
944 * doc/invoke.texi (armv7-r): Add two new vfp options.
945 (nofp.dp): Add cortex-r7 and cortex-r8 to the list of targets that
948 2018-11-29 Alan Modra <amodra@gmail.com>
950 * config/rs6000/rs6000.c (rs6000_emit_move): Disable long
951 double split for targets other than Darwin.
953 2018-11-29 Alan Modra <amodra@gmail.com>
955 * config/rs6000/rs6000.md (movdi_internal64): Correct lengths.
957 2018-11-29 Alan Modra <amodra@gmail.com>
959 * config/rs6000/predicates.md (easy_fp_constant): Remove code
960 dealing with integers in const_double. Assert on unexpected
961 modes. Delete superfluous ABI_V4 test.
962 * config/rs6000/rs6000.c (rs6000_emit_move): Comment fixes.
963 Don't call easy_fp_constant for Pmode.
964 (rs6000_hash_constant): Remove code dealing with integers in
966 (rs6000_legitimate_constant_p): Likewise.
967 (output_toc): Formatting, use CONST_DOUBLE_P.
969 2018-11-29 Alan Modra <amodra@gmail.com>
971 * config/rs6000/constraints.md (G, H): Comment on purpose of
972 constraint. Correct mode comments and "G" mode test.
973 * config/rs6000/rs6000.md (movdi_internal32): Remove "GH" from
974 alternative handling "F". Add length attr. Formatting.
975 (mov<mode>_softfloat <FMOVE128>): Delete "GH" from
976 alternative, and move "F" to separate alternative. Correct
979 2018-11-29 Alan Modra <amodra@gmail.com>
981 * config/rs6000/predicates.md (easy_vector_constant): Don't call
984 2018-11-29 Alan Modra <amodra@gmail.com>
986 * config.in (HAVE_AS_PLTSEQ): Add.
987 * config/rs6000/predicates.md (indirect_call_operand): New.
988 * config/rs6000/rs6000-protos.h (rs6000_pltseq_template),
989 (rs6000_sibcall_sysv): Declare.
990 * config/rs6000/rs6000.c (init_cumulative_args): Set cookie
991 CALL_LONG for -fno-plt.
992 (print_operand <T, z, 0>): Handle UNSPEC_PLTSEQ.
993 (rs6000_indirect_call_template_1): Emit .reloc directives for
995 (rs6000_pltseq_template): New function.
996 (rs6000_longcall_ref): Add arg parameter. Use PLT16 insns if
997 relocs supported by assembler. Move SYMBOL_REF test to callers.
998 (rs6000_call_aix): Adjust rs6000_longcall_ref call. Package
999 insns in UNSPEC_PLTSEQ, preserving original func_desc.
1000 (rs6000_call_sysv): Likewise.
1001 (rs6000_sibcall_sysv): New function.
1002 * config/rs6000/rs6000.h (HAVE_AS_PLTSEQ): Provide default.
1003 * config/rs6000/rs6000.md (UNSPEC_PLTSEQ, UNSPEC_PLT16_HA,
1004 UNSPEC_PLT16_LO): New.
1005 (pltseq_tocsave, pltseq_plt16_ha, pltseq_plt16_lo, pltseq_mtctr): New.
1006 (call_indirect_nonlocal_sysv): Don't differentiate zero from non-zero
1007 cookie in constraints. Test explicitly for flags in length attr.
1008 Handle unspec operand 1.
1009 (call_value_indirect_nonlocal_sysv): Likewise.
1010 (call_indirect_aix, call_value_indirect_aix): Handle unspec operand 1.
1011 (call_indirect_elfv2, call_value_indirect_elfv2): Likewise.
1012 (sibcall, sibcall_value): Use rs6000_sibcall_sysv.
1013 (sibcall_indirect_nonlocal_sysv): New pattern.
1014 (sibcall_value_indirect_nonlocal_sysv): Likewise.
1015 (sibcall_nonlocal_sysv, sibcall_value_nonlocal_sysv): Remove indirect
1017 * configure.ac: Check for gas plt sequence marker support.
1018 * configure: Regenerate.
1020 2018-11-29 Alan Modra <amodra@gmail.com>
1022 * config/rs6000/predicates.md (unspec_tls): New.
1023 * config/rs6000/rs6000-protos.h (rs6000_call_template),
1024 (rs6000_sibcall_template): Update prototype.
1025 (rs6000_longcall_ref): Delete.
1026 (rs6000_call_sysv): Declare.
1027 * config/rs6000/rs6000.c (edit_tls_call_insn): New function.
1028 (global_tlsarg): New variable.
1029 (rs6000_legitimize_tls_address): Rewrite __tls_get_addr call
1031 (print_operand): Extract UNSPEC_TLSGD address operand.
1032 (rs6000_call_template, rs6000_sibcall_template): Remove arg
1033 parameter, extract from second call operand instead.
1034 (rs6000_longcall_ref): Make static, localize vars.
1035 (rs6000_call_aix): Rename parameter to reflect new usage. Take
1036 tlsarg from global_tlsarg. Don't create unused rtl or nop insns.
1037 (rs6000_sibcall_aix): Rename parameter to reflect new usage. Take
1038 tlsarg from global_tlsarg.
1039 (rs6000_call_sysv): New function.
1040 * config/rs6000/rs6000.md: Adjust rs6000_call_template and
1041 rs6000_sibcall_template throughout.
1042 (tls_gd_aix, tls_gd_sysv, tls_gd_call_aix, tls_gd_call_sysv): Delete.
1043 (tls_ld_aix, tls_ld_sysv, tls_ld_call_aix, tls_ld_call_sysv): Delete.
1044 (tls_gdld_nomark): New insn.
1045 (tls_gd): Swap operand order. Simplify mode selection.
1046 (tls_gd_high, tls_gd_low): Swap operand order.
1047 (tls_ld): Remove const_int 0 vector element from UNSPEC_TLSLD.
1048 Simplify mode selection.
1049 (tls_ld_high, tls_ld_low): Similarly adjust UNSPEC_TLSLD.
1050 (call, call_value): Don't assert for second call operand.
1051 Use rs6000_call_sysv.
1053 2018-11-29 Alan Modra <amodra@gmail.com>
1055 * config/rs6000/darwin.md (call_indirect_nonlocal_darwin64),
1056 (call_nonlocal_darwin64, call_value_indirect_nonlocal_darwin64),
1057 (call_value_nonlocal_darwin64): Remove constraints from second call
1058 arg, the rounded_stack_size_rtx arg.
1059 * config/rs6000/rs6000.md (tls_gd_aix, tls_gd_sysv, tls_gd_call_aix),
1060 (tls_gd_call_sysv, tls_ld_aix, tls_ld_sysv, tls_ld_call_aix),
1061 (tls_ld_call_sysv, call_local32, call_local64, call_value_local32),
1062 (call_value_local64, call_indirect_nonlocal_sysv),
1063 (call_nonlocal_sysv, call_nonlocal_sysv_secure),
1064 (call_value_indirect_nonlocal_sysv, call_value_nonlocal_sysv),
1065 (call_value_nonlocal_sysv_secure, call_local_aix),
1066 (call_value_local_aix, call_nonlocal_aix, call_value_nonlocal_aix),
1067 (call_indirect_aix, call_value_indirect_aix, call_indirect_elfv2),
1068 (call_value_indirect_elfv2, sibcall_local32, sibcall_local64),
1069 (sibcall_value_local32, sibcall_value_local64, sibcall_aix),
1070 (sibcall_value_aix): Likewise.
1072 2018-11-29 Alan Modra <amodra@gmail.com>
1074 * config/rs6000/rs6000.md (TLSmode): Delete mode iterator. Replace
1075 with P throughout except for call mems which should use SI.
1076 (tls_abi_suffix, tls_sysv_suffix, tls_insn_suffix): Delete mode
1077 attributes. Replace with bits, mode and ptrload respectively.
1079 2018-11-29 Alan Modra <amodra@gmail.com>
1081 * config/rs6000/rs6000-protos.h (rs6000_indirect_call_template),
1082 (rs6000_indirect_sibcall_template): Declare.
1083 * config/rs6000/rs6000.c (rs6000_indirect_call_template_1),
1084 (rs6000_indirect_call_template, rs6000_indirect_sibcall_template):
1086 * config/rs6000/rs6000.md (call_indirect_nonlocal_sysv),
1087 (call_value_indirect_nonlocal_sysv, sibcall_nonlocal_sysv),
1088 (call_indirect_aix, call_value_indirect_aix): Use
1089 rs6000_indirect_call_template and rs6000_indirect_sibcall_template.
1090 call_indirect_elfv2, call_value_indirect_elfv2): Likewise, and
1091 handle both speculation and non-speculation cases.
1092 (call_indirect_aix_nospec, call_value_indirect_aix_nospec): Delete.
1093 (call_indirect_elfv2_nospec, call_value_indirect_elfv2_nospec): Delete.
1095 2018-11-29 Alan Modra <amodra@gmail.com>
1097 * config/rs6000/rs6000-protos.h (rs6000_call_template): Declare.
1098 (rs6000_sibcall_template): Declare.
1099 (macho_call_template): Rename from output_call.
1100 * config/rs6000/rs6000.c (rs6000_call_template_1): New function.
1101 (rs6000_call_template, rs6000_sibcall_template): Likewise.
1102 (macho_call_template): Rename from output_call.
1103 * config/rs6000/rs6000.md (tls_gd_aix, tls_gd_sysv),
1104 (tls_gd_call_aix, tls_gd_call_sysv, tls_ld_aix, tls_ld_sysv),
1105 (tls_ld_call_aix, tls_ld_call_sysv, call_nonlocal_sysv),
1106 (call_nonlocal_sysv_secure, call_value_nonlocal_sysv),
1107 (call_value_nonlocal_sysv_secure, call_nonlocal_aix),
1108 (call_value_nonlocal_aix): Use rs6000_call_template and update
1109 occurrences of output_call to macho_call_template.
1110 (sibcall_nonlocal_sysv, sibcall_value_nonlocal_sysv, sibcall_aix),
1111 (sibcall_value_aix): Use rs6000_sibcall_template.
1113 2018-11-28 Aaron Sawdey <acsawdey@linux.ibm.com>
1115 * config/rs6000/rs6000-string.c (expand_block_clear): Change how
1116 we determine if unaligned vsx is ok.
1118 2018-11-28 Jan Hubicka <jh@suse.cz>
1120 * predict.c (determine_unlikely_bbs): Forward declare; also determine
1121 edges that are always known to be taken because it is only likely
1122 edge out of given BB.
1123 (tree_estimate_probability): Call before profile guessing to get bit
1126 2018-11-28 Jan Hubicka <jh@suse.cz>
1128 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Handle
1129 profile_probability::always better.
1131 2018-11-28 Jan Hubicka <jh@suse.cz>
1133 * profile-count.h (profile_count::split): Give better result when
1134 splitting profile_probability::always.
1136 2018-11-28 Vladimir Makarov <vmakarov@redhat.com>
1139 * ira-costs.c (scan_one_insn): Process subregs when updating costs
1140 for pseudos and allocnos from insn.
1142 2018-11-28 David Edelsohn <dje.gcc@gmail.com>
1144 * config/rs6000/aix72.h: Update to match aix71.h changes.
1146 2018-11-28 Jeff Law <law@redhat.com>
1148 * config/h8300/constraints.md: Add "C" constraint for call insns.
1149 * config/h8300/h8300.md (call, call_value): Turn into a define_expand
1150 and define_insn pair. Move invalid call targets into a register in
1151 the expander and fix constraints in the matching pattern.
1152 * config/h8300/predicates.md (call_expander_operand): Renamed from
1153 call_insn_operand. Reject things we shouldn't be trying to handle.
1154 (call_insn_operand): New predicate for use by the call/call_value
1156 (small_call_insn_operand): Update appropriately.
1158 2018-11-28 Sam Tebbs <sam.tebbs@arm.com>
1160 * config/aarch64/aarch64.c (aarch64_process_target_attr): Replace
1161 calls to strtok with strtok_r.
1163 2018-11-28 Richard Biener <rguenther@suse.de>
1165 PR tree-optimization/88223
1166 * tree-ssa-sccvn.c (vn_reference_lookup_3): When skipping
1167 over a stored-same value may-alias store make sure to consider
1168 partial overlaps which are valid when TBAA reasonings do not
1169 apply and byte-granular overlaps are possible at all.
1171 2018-11-28 Richard Biener <rguenther@suse.de>
1173 PR tree-optimization/88217
1174 * vr-values.c (vr_values::extract_range_from_phi_node): Make
1175 sure to handle results > +INF and < -INF correctly when
1176 trying to drop down to +INF - 1 or -INF + 1.
1178 2018-11-28 Alan Modra <amodra@gmail.com>
1180 * xcoffout.c (do_block): Signed/unsigned warning fix.
1182 2018-11-28 Richard Biener <rguenther@suse.de>
1184 PR tree-optimization/79351
1185 * tree-ssa-sccvn.c (vn_reference_lookup_3): For assignments from
1186 empty CONSTRUCTORs ensure the store is at a constant position.
1188 2018-11-28 Richard Biener <rguenther@suse.de>
1190 PR tree-optimization/88229
1191 * tree-ssa.c (non_rewritable_mem_ref_base): Check DECL_SIZE_UNIT
1192 is an INTEGER_CST before accessing it so.
1194 2018-11-28 Sam Tebbs <sam.tebbs@arm.com>
1196 * doc/invoke.texi (-mtune=): Remove obsolete CPU names.
1198 2018-11-28 Jakub Jelinek <jakub@redhat.com>
1201 * config/i386/i386.c (ix86_expand_sse_movcc): Handle DFmode and
1202 SFmode using sse4_1_blendvs[sd] with TARGET_SSE4_1. Formatting fixes.
1203 * config/i386/sse.md (sse4_1_blendv<ssemodesuffix>): New pattern.
1205 2018-11-27 Martin Liska <mliska@suse.cz>
1207 * gcov.c (generate_results): Append current_working_directory
1210 2018-11-27 Mike Gulick <mgulick@mathworks.com>
1212 PR preprocessor/83173
1213 * input.c (dump_location_info): Dump reason and included_from
1214 fields from line_map_ordinary struct. Fix indentation when
1215 location > 5 digits.
1216 * diagnostic-show-locus.c (num_digits, num_digits): Move to
1217 diagnostic.c to allow it to be utilized by input.c.
1218 * diagnostic.c (num_digits, selftest::test_num_digits): Moved
1220 (selftest::diagnostic_c_tests): Run selftest::test_num_digits.
1221 * diagnostic.h (num_digits): Add extern definition.
1223 2018-11-27 Fredrik Noring <noring@nocrew.org>
1225 * config/mips/mips.c (mips_reorg_process_insns)
1226 (mips_option_override): Handle `-mfix-r5900'.
1227 * config/mips/mips.h (ASM_SPEC): Add `mfix-r5900' and
1229 * config/mips/mips.opt (mfix-r5900): New option.
1230 * doc/invoke.texi: Document the `r5900' processor name, and
1231 `-mfix-r5900' and `-mno-fix-r5900' options.
1233 2018-11-27 Jakub Jelinek <jakub@redhat.com>
1236 * config/rs6000/rs6000.c (print_operand) <case 'D'>: Use
1237 output_operand_lossage instead of gcc_assert.
1238 <case 't'>: Likewise.
1239 <case 'z'>: Likewise.
1240 <case 'V'>: Use output_operand_lossage instead of gcc_unreachable.
1242 2018-11-27 Jeff Law <law@redhat.com>
1244 * config/riscv/riscv (riscv_block_mvoe_straight): Use RETURN_BEGIN
1245 in call to move_by_pieces.
1247 * config/sh/sh-mem.c (expand_block_move): Use RETURN_BEGIN in call
1250 * config/lm32/lm32.c (lm32_block_move_inline): Use RETURN_BEGIN in
1251 call to move_by_pieces.
1253 * config/mips/mips.c (mips_block_move_straight): Use RETURN_BEGIN
1254 in call to move_by_pieces.
1256 * config/microblaze/microblaze.c (microblaze_block_move_straight): Use
1257 RETURN_BEGIN in call to move_by_pieces.
1258 (microblaze_expand_block_move): Likewise.
1260 2018-11-27 Tamar Christina <tamar.christina@arm.com>
1262 * doc/md.texi (xorsign): Document it.
1264 2018-11-27 Martin Liska <mliska@suse.cz>
1266 * config/gnu-user.h (TARGET_F951_OPTIONS): New.
1267 * gcc.c (find_fortran_preinclude_file): New function
1268 to handle Fortran pre-include.
1270 2018-11-27 Martin Liska <mliska@suse.cz>
1272 * asan.c (asan_emit_stack_protection): Use new enum values
1273 instead of int constants.
1274 * builtins.c (expand_builtin_memory_copy_args): Replace int
1275 type with memop_ret enum type.
1276 (expand_builtin_mempcpy_args): Likewise.
1277 (expand_builtin_memcpy): Use new enum values
1278 instead of int constants. Likewise.
1279 (expand_builtin_mempcpy): Likewise.
1280 (expand_movstr): Likewise.
1281 (expand_builtin_strcpy_args): Likewise.
1282 (expand_builtin_stpcpy_1): Likewise.
1283 (expand_builtin_strncpy): Likewise.
1284 (expand_builtin_memset_args): Likewise.
1285 * expr.c (move_by_pieces_d::finish_endp): Rename to ...
1286 (move_by_pieces_d::finish_retmode): ... this.
1287 (move_by_pieces): Change last argument type to memop_ret.
1288 (store_by_pieces): Use new enum values
1289 instead of int constants.
1290 (emit_block_move_hints): Likewise.
1291 (emit_push_insn): Likewise.
1292 (store_expr): Likewise.
1293 * expr.h (store_by_pieces): Change int to newly added enum
1295 * rtl.h (enum memop_ret): Define.
1296 (move_by_pieces): Use the enum type.
1298 2018-11-27 Alan Modra <amodra@gmail.com>
1300 * config/rs6000/aix71.h (ASM_SPEC): Don't select default -maix64
1302 (ASM_CPU_SPEC): Do so here. Rewrite using if .. else if .. specs
1303 form. Error on missing -mcpu case.
1304 * config/rs6000/driver-rs6000.c (asm_names <_AIX>): Update NULL case.
1305 (asm_names <!_AIX>): Add missing cpus. Update NULL case. Apply
1306 PR63177 fix for -mcpu=power8 and -mcpu=powerpc64le.
1307 * config/rs6000/rs6000.h (ASM_CPU_SPEC): Rewrite using if ..
1308 else if .. specs form. Error on missing -mcpu case. Don't output
1309 duplicate -maltivec. Apply PR63177 fix for -mcpu=powerpc64le.
1311 2018-11-26 David Malcolm <dmalcolm@redhat.com>
1313 * dump-context.h (dump_context::dump_loc): Convert 1st param from
1314 dump_flags_t to const dump_metadata_t &. Convert 2nd param from
1315 const dump_location_t & to const dump_user_location_t &.
1316 (dump_context::dump_loc_immediate): Convert 2nd param from
1317 const dump_location_t & to const dump_user_location_t &.
1318 (dump_context::dump_gimple_stmt): Convert 1st param from
1319 dump_flags_t to const dump_metadata_t &.
1320 (dump_context::void dump_gimple_stmt_loc): Likewise; convert
1321 2nd param from const dump_location_t & to
1322 const dump_user_location_t &.
1323 (dump_context::dump_gimple_expr): Convert 1st param from
1324 dump_flags_t to const dump_metadata_t &.
1325 (dump_context::dump_gimple_expr_loc): Likewise; convert
1326 2nd param from const dump_location_t & to
1327 const dump_user_location_t &.
1328 (dump_context::dump_generic_expr): Convert 1st param from
1329 dump_flags_t to const dump_metadata_t &.
1330 (dump_context::dump_generic_expr_loc): Likewise; convert
1331 2nd param from const dump_location_t & to
1332 const dump_user_location_t &.
1333 (dump_context::dump_printf_va): Convert 1st param from
1334 dump_flags_t to const dump_metadata_t &.
1335 (dump_context::dump_printf_loc_va): Likewise; convert
1336 2nd param from const dump_location_t & to
1337 const dump_user_location_t &.
1338 (dump_context::dump_dec): Convert 1st param from
1339 dump_flags_t to const dump_metadata_t &.
1340 (dump_context::dump_symtab_node): Likewise.
1341 (dump_context::begin_scope): Split out 2nd param into
1342 user and impl locations.
1343 (dump_context::ensure_pending_optinfo): Add metadata param.
1344 (dump_context::begin_next_optinfo): Replace dump_location_t param
1345 with metadata and user location.
1346 * dumpfile.c (dump_context::dump_loc): Convert 1st param from
1347 dump_flags_t to const dump_metadata_t &. Convert 2nd param from
1348 const dump_location_t & to const dump_user_location_t &.
1349 (dump_context::dump_loc_immediate): Convert 2nd param from
1350 const dump_location_t & to const dump_user_location_t &.
1351 (dump_context::dump_gimple_stmt): Convert 1st param from
1352 dump_flags_t to const dump_metadata_t &.
1353 (dump_context::void dump_gimple_stmt_loc): Likewise; convert
1354 2nd param from const dump_location_t & to
1355 const dump_user_location_t &.
1356 (dump_context::dump_gimple_expr): Convert 1st param from
1357 dump_flags_t to const dump_metadata_t &.
1358 (dump_context::dump_gimple_expr_loc): Likewise; convert
1359 2nd param from const dump_location_t & to
1360 const dump_user_location_t &.
1361 (dump_context::dump_generic_expr): Convert 1st param from
1362 dump_flags_t to const dump_metadata_t &.
1363 (dump_context::dump_generic_expr_loc): Likewise; convert
1364 2nd param from const dump_location_t & to
1365 const dump_user_location_t &.
1366 (dump_context::dump_printf_va): Convert 1st param from
1367 dump_flags_t to const dump_metadata_t &.
1368 (dump_context::dump_printf_loc_va): Likewise; convert
1369 2nd param from const dump_location_t & to
1370 const dump_user_location_t &.
1371 (dump_context::dump_dec): Convert 1st param from
1372 dump_flags_t to const dump_metadata_t &.
1373 (dump_context::dump_symtab_node): Likewise.
1374 (dump_context::begin_scope): Split out 2nd param into
1375 user and impl locations.
1376 (dump_context::ensure_pending_optinfo): Add metadata param.
1377 (dump_context::begin_next_optinfo): Replace dump_location_t param
1378 with metadata and user location.
1379 (dump_gimple_stmt): Convert 1st param from dump_flags_t to
1380 const dump_metadata_t &.
1381 (dump_gimple_stmt_loc): Likewise; convert 2nd param from
1382 const dump_location_t & to const dump_user_location_t &.
1383 (dump_gimple_expr): Convert 1st param from dump_flags_t to
1384 const dump_metadata_t &.
1385 (dump_gimple_expr_loc): Likewise; convert 2nd param from
1386 const dump_location_t & to const dump_user_location_t &.
1387 (dump_generic_expr): Convert 1st param from dump_flags_t to
1388 const dump_metadata_t &.
1389 (dump_generic_expr_loc): Likewise; convert 2nd param from
1390 const dump_location_t & to const dump_user_location_t &.
1391 (dump_printf): Convert 1st param from dump_flags_t to
1392 const dump_metadata_t &.
1393 (dump_printf_loc): Likewise; convert 2nd param from
1394 const dump_location_t & to const dump_user_location_t &.
1395 (dump_dec): Convert 1st param from dump_flags_t to
1396 const dump_metadata_t &.
1397 (dump_symtab_node): Likewise.
1398 (dump_begin_scope): Split out 2nd param into user and impl
1400 (selftest::assert_impl_location_eq): New function.
1401 (ASSERT_IMPL_LOCATION_EQ): New macro.
1402 (selftest::test_impl_location): Update to use
1403 ASSERT_IMPL_LOCATION_EQ.
1404 (selftest::test_capture_of_dump_calls): Convert "loc" to
1405 dump_user_location_t. Add ASSERT_IMPL_LOCATION_EQ throughout,
1406 verifying line numbers of dump emissions.
1407 * dumpfile.h (class dump_metadata_t): New class.
1408 (dump_printf): Convert 1st param from dump_flags_t to
1409 const dump_metadata_t &.
1410 (dump_printf_loc): Likewise; convert 2nd param from
1411 const dump_location_t & to const dump_user_location_t &.
1412 (dump_generic_expr_loc): Likewise.
1413 (dump_generic_expr): Convert 1st param from dump_flags_t to
1414 const dump_metadata_t &.
1415 (dump_gimple_stmt_loc): Likewise; convert 2nd param from
1416 const dump_location_t & to const dump_user_location_t &.
1417 (dump_gimple_stmt): Convert 1st param from dump_flags_t to
1418 const dump_metadata_t &.
1419 (dump_gimple_expr_loc): Likewise; convert 2nd param from
1420 const dump_location_t & to const dump_user_location_t &.
1421 (dump_gimple_expr): Convert 1st param from dump_flags_t to
1422 const dump_metadata_t &.
1423 (dump_symtab_node): Likewise.
1424 (dump_dec): Likewise.
1425 (dump_begin_scope): Split out 2nd param into user and impl
1427 (auto_dump_scope::auto_dump_scope): Split "loc" param into a user
1428 location and impl_location, and capture the impl_location.
1429 (AUTO_DUMP_SCOPE): Rename param from LOC to USER_LOC.
1430 * loop-unroll.c (report_unroll): Update for changes to
1431 dump_printf_loc and dump_printf.
1432 * opt-problem.cc (opt_problem::opt_problem): Update for change to
1434 * optinfo-emit-json.cc
1435 (selftest::test_building_json_from_dump_calls): Convert "loc" from
1436 dump_location_t to dump_user_location_t.
1437 * optinfo.cc (optinfo::emit_for_opt_problem): Update for change in
1439 * profile.c (compute_branch_probabilities): Update for change to
1441 * selftest.h (ASSERT_STR_CONTAINS_AT): New macro.
1442 * tree-vect-slp.c (vect_print_slp_tree): Update for change to
1445 2018-11-27 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1447 * doc/sourcebuild.texi: Document check_effective_target_int_eq_float
1448 and check_effective_target_ptr_eq_long.
1450 2018-11-27 Alan Modra <amodra@gmail.com>
1452 * config.gcc (powerpc*-*-freebsd*, powerpc-*-netbsd*),
1453 (powerpc-*-eabisimaltivec*, powerpc-*-eabisim*, powerpc-*-elf*),
1454 (powerpc-*-eabialtivec*, powerpc-*-eabi*, powerpc-*-rtems*),
1455 (powerpc-wrs-vxworks*, powerpc-*-lynxos*, powerpcle-*-elf*),
1456 (powerpcle-*-eabisim*, powerpcle-*-eabi*): Add gnu-user.h to tm_file.
1457 * config/rs6000/freebsd.h (CPLUSPLUS_CPP_SPEC),
1458 (LINK_GCC_C_SEQUENCE_SPEC): Undef.
1459 (ASM_APP_ON, ASM_APP_OFF): Don't define.
1460 * config/rs6000/freebsd64.h (ASM_APP_ON, ASM_APP_OFF): Don't define.
1461 * config/rs6000/lynx.h (ASM_APP_ON, ASM_APP_OFF): Don't define.
1462 * config/rs6000/linux64.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
1463 * config/rs6000/netbsd.h (CPLUSPLUS_CPP_SPEC),
1464 (LINK_GCC_C_SEQUENCE_SPEC): Undef.
1465 * config/rs6000/rtems.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
1466 * config/rs6000/sysv4.h (GNU_USER_TARGET_CRTI): Redefine.
1467 (GNU_USER_TARGET_CRTN): Redefine.
1468 (CC1_SPEC): Use GNU_USER_TARGET_CC1_SPEC.
1469 (LIB_LINUX_SPEC): Use GNU_USER_TARGET_LIB_SPEC.
1470 (CRTOFFLOADBEGIN, CRTOFFLOADEND): Don't define.
1471 (STARTFILE_LINUX_SPEC): Define as GNU_USER_TARGET_STARTFILE_SPEC.
1472 (ENDFILE_LINUX_SPEC): Define as GNU_USER_TARGET_ENDFILE_SPEC.
1473 (UCLIBC_DYNAMIC_LINKER, CHOOSE_DYNAMIC_LINKER): Don't define.
1474 (LINK_EH_SPEC): Don't define.
1476 2018-11-26 Martin Sebor <msebor@redhat.com>
1479 * expr.c (string_constant): Handle top-level decls of all character
1480 types and subobjects of narrow character type.
1482 2018-11-27 Alan Modra <amodra@gmail.com>
1484 * config.gcc (powerpc*-*-linux*): Add linux.h to tm_file.
1485 * config/rs6000/linux.h (TARGET_OS_CPP_BUILTINS): Use
1486 GNU_USER_TARGET_OS_CPP_BUILTINS.
1487 (RS6000_ABI_NAME): Define.
1488 * config/rs6000/linux64.h (TARGET_OS_CPP_BUILTINS): Use
1489 GNU_USER_TARGET_OS_CPP_BUILTINS.
1490 (MUSL_DYNAMIC_LINKER32): Undef before defining.
1491 (UCLIBC_DYNAMIC_LINKER32, UCLIBC_DYNAMIC_LINKER64): Don't define.
1492 (CHOOSE_DYNAMIC_LINKER): Don't define.
1493 (GNU_USER_DYNAMIC_LINKER32, GNU_USER_DYNAMIC_LINKER64): Don't define.
1494 * config/rs6000/sysv4.h (MUSL_DYNAMIC_LINKER): Undef before defining.
1495 (CHOOSE_DYNAMIC_LINKER, GNU_USER_DYNAMIC_LINKER): Only define when
1496 not already defined.
1497 (CPP_OS_LINUX_SPEC): Remove defines and asserts handled by
1498 TARGET_OS_CPP_BUILTINS.
1500 2018-11-26 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1502 * doc/sourcebuild.texi: Document check_effective_target_newlib_nano_io.
1504 2018-11-26 Jakub Jelinek <jakub@redhat.com>
1507 * config/i386/i386.c (def_builtin2): If tcode == VOID_FTYPE_UINT64
1508 and !TARGET_64BIT, return NULL_TREE.
1511 * dwarf2out.c (secname_for_decl): For functions with
1512 DECL_SECTION_NAME if in_cold_section_p, try to return
1513 current_function_section's name if it is a named section.
1515 2018-11-26 Maya Rashish <coypu@sdf.org>
1518 * config.host (*-*-netbsd*): Use x-netbsd and host-netbsd.o.
1519 * config/x-netbsd: New file.
1520 * config/host-netbsd: New file.
1522 2018-11-26 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1524 * doc/sourcebuild.texi: Document check_effective_target_size20plus.
1525 Clarify documentation for check_effective_target_size32plus.
1527 2018-11-26 Aaron Sawdey <acsawdey@linux.ibm.com>
1529 * config/rs6000/rs6000-string.c (expand_cmp_vec_sequence): Rename
1530 and modify expand_strncmp_vec_sequence.
1531 (emit_final_compare_vec): Rename and modify emit_final_str_compare_vec.
1532 (generate_6432_conversion): New function.
1533 (expand_block_compare): Add support for vsx.
1534 (expand_block_compare_gpr): New function.
1535 * config/rs6000/rs6000.opt (rs6000_block_compare_inline_limit): Increase
1536 default limit to 63 because of more compact vsx code.
1538 2018-11-26 Uros Bizjak <ubizjak@gmail.com>
1541 * config/i386/i386.c (dbx_register_map): Use IGNORED_DWARF_REGNUM
1542 for registers for which no debug information can be generated.
1543 Use INVALID_REGNUM for non-existent registers.
1544 (dbx64_register_map): Ditto.
1545 (svr4_dbx_register_map): Ditto.
1547 2018-11-26 Sandra Loosemore <sandra@codesourcery.com>
1550 * doc/invoke.texi (Optimize Options): Document that
1551 -freorder-functions also uses "hot" and "cold" attributes.
1553 2018-11-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1555 * config/aarch64/aarch64.c (aarch64_mangle_type): Fix typo in comment.
1557 2018-11-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1559 * config/arm/types.md (mul64): Rename to...
1560 (widen_mul64): ... This.
1561 * config/arm/arm-generic.md: Rename mul64 to widen_mul64.
1562 * config/arm/cortex-a15.md: Likewise.
1563 * config/arm/cortex-a5.md: Likewise.
1564 * config/arm/cortex-a53.md: Likewise.
1565 * config/arm/cortex-a57.md: Likewise.
1566 * config/arm/cortex-a7.md: Likewise.
1567 * config/arm/cortex-m4.md: Likewise.
1568 * config/arm/exynos-m1.md: Likewise.
1569 * config/arm/marvell-pj4.md: Likewise.
1570 * config/arm/xgene1.md: Likewise.
1572 2018-11-26 Richard Biener <rguenther@suse.de>
1574 PR tree-optimization/88182
1575 * tree-vect-loop.c (vectorizable_reduction): Pick up single
1576 correct reduc_def_info.
1577 * tree-vect-slp.c (vect_analyze_slp_instance): Set
1578 STMT_VINFO_REDUC_DEF of the first stmt.
1580 2018-11-26 Andreas Krebbel <krebbel@linux.ibm.com>
1582 * doc/invoke.texi: Document z14/arch12 -march option.
1584 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
1587 * doc/invoke.texi (Option Summary): Add -Wmissing-noreturn.
1588 (Warning Options): Likewise.
1590 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
1593 * doc/extend.texi (Common Function Attributes): Clarify that
1594 functions with "const" attribute can read const global variables.
1596 2018-11-26 Alan Modra <amodra@gmail.com>
1598 * config/rs6000/rs6000.h (OBJECT_PEF, TARGET_MACOS): Delete.
1600 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
1603 * doc/extend.texi (Common Variable Attributes): Use preferred
1604 placement of type attributes in examples, plus whitespace fixes.
1605 (Type Attributes): Clarify why placement of attributes
1606 immediately after struct/union/enum keyword is preferred.
1607 (Common Type Attributes): Use preferred placement of type
1608 attributes in examples, plus more whitespace fixes.
1610 2018-11-25 Paul Koning <ni1d@arrl.net>
1612 * config/pdp11/pdp11.h (TARGET_HAS_NO_HW_DIVIDE): Define.
1614 2018-11-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1616 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
1617 attributes): Document d_runtime.
1619 2018-11-25 Vladimir Makarov <vmakarov@redhat.com>
1622 * ira-costs.c (record_operand_costs): Initiate register move cost
1625 2018-11-23 Jeff Law <law@redhat.com>
1627 PR rtl-optimization/87468
1628 * tree-ssa-threadupdate.c (create_block_for_threading): Clear
1629 EDGE_IGNORE on all outgoing edges of the duplicate block.
1631 2018-11-23 Vladimir Makarov <vmakarov@redhat.com>
1634 * ira-costs.c (record_operand_costs): Use bigger hard reg class if
1635 its mode does not fit to the original class.
1637 2018-11-23 Martin Sebor <msebor@redhat.com>
1639 PR tree-optimization/87756
1640 * expr.c (string_constant): Synthesize a string literal from
1641 the address of a constant character.
1642 * tree.c (build_string_literal): Add an argument.
1643 * tree.h (build_string_literal): Same.
1645 2018-11-23 Christoph Muellner <christoph.muellner@theobroma-systems.com>
1647 * config/aarch64/aarch64-cores.def: Define emag.
1648 * config/aarch64/aarch64-tune.md: Regenerated with emag.
1649 * config/aarch64/aarch64.c (emag_tunings): New struct.
1650 * doc/invoke.texi: Document mtune value.
1652 2018-11-23 Christoph Muellner <christoph.muellner@theobroma-systems.com>
1653 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
1655 * config/aarch64/aarch64.c (xgene1_tunings): Optimize Xgene1 tunings
1657 * config/aarch64/aarch64.c (xgene1_tunings): Add Xgene1 specific
1659 * config/aarch64/aarch64.c (xgene1_addrcost_table): Correct the post-modify
1661 * config/arm/aarch-cost-tables.h (xgene1_extra_costs): Update the cost table
1664 2018-11-23 Richard Biener <rguenther@suse.de>
1666 PR tree-optimization/88149
1667 * tree-vect-slp.c (vect_slp_analyze_node_operations): Detect
1668 the case where there are two different def types for the
1669 same operand at different operand position in the same stmt.
1671 2018-23-11 Mihail Ionescu <mihail.ionescu@arm.com>
1673 * config/arm/arm.c (arm_expand_compare_and_swap): Simplify and call
1674 gen_atomic_compare_swap_1.
1675 (arm_evpc_neon_vuzp): Likewise gen_neon_vuzp_internal.
1676 (arm_evpc_neon_vtrn): Likewise gen_neon_vtrn_internal.
1677 (arm_evpc_neon_vext): Likewise gen_neon_vext_internal.
1678 (arm_evpc_neon_vzip): Likewise gen_neon_vzip_internal.
1679 (arm_evpc_neon_vrev): Replace the function pointer and simplify the mode
1681 * config/arm/arm.md (neon_vext<mode>),
1682 (neon_vrev64<mode>, neon_vrev32<mode>),
1683 (neon_vrev16<mode>, neon_vtrn<mode>_internal),
1684 (neon_vzip<mode>_internal, neon_vuzp<mode>_internal): Add an '@'character
1685 before the pattern name.
1686 * config/arm/sync.md:
1687 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1),
1688 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): Likewise.
1690 2018-11-23 Jakub Jelinek <jakub@redhat.com>
1692 PR tree-optimization/86614
1693 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return early
1694 if TREE_NO_WARNING is set on ref.ref.
1696 2018-11-22 Sandra Loosemore <sandra@codesourcery.com>
1697 Alan Coopersmith <alan.coopersmith@oracle.com>
1700 * doc/extend.texi (Designated Inits): Clarify handling of multiple
1701 initializers for unions.
1703 2018-11-22 Jan Hubicka <jh@suse.cz>
1706 * ipa-devirt.c (type_variants_equivalent_p): Drop warn and warned
1707 parameters; do not warn here.
1708 (odr_subtypes_equivalent_p): Likewise.
1709 (warn_odr): Fix typo.
1710 (warn_types_mismatch): Do not output confused warnings on integer types.
1711 (odr_types_equivalent_p): Update.
1713 2018-11-22 Jakub Jelinek <jakub@redhat.com>
1715 * config/i386/i386.c (ix86_option_override_internal,
1716 ix86_can_inline_p, classify_argument, construct_container,
1717 ix86_expand_prologue, ix86_expand_split_stack_prologue,
1718 ix86_expand_carry_flag_compare, expand_set_or_movmem_via_loop,
1719 expand_setmem_epilogue_via_loop, promote_duplicated_reg,
1720 ix86_expand_set_or_movmem, ix86_init_builtins_va_builtins_abi):
1723 * config/i386/i386.c (ix86_option_override_internal): For
1724 stack_protector_guard related options, use opts_set->x_ instead
1725 of global_options_set. and prefix options with opts->x_ . Move
1726 defaults for offset and reg into else block.
1728 2018-11-22 Uros Bizjak <ubizjak@gmail.com>
1730 * config/i386/i386.c (ix86_check_avx_upper_register):
1731 Return true for all SSE registers with mode bitsize > 128.
1733 2018-11-22 Vladimir Makarov <vmakarov@redhat.com>
1735 PR rtl-optimization/87718
1736 * ira-costs.c: Remove trailing white-spaces.
1737 (record_operand_costs): Add a special treatment for moves
1738 involving a hard register.
1740 2018-11-22 Uros Bizjak <ubizjak@gmail.com>
1742 * config/i386/i386.c (ix86_avx_emit_vzeroupper): Remove.
1743 (ix86_emit_mode_set) <case AVX_U128>: Emit vzeroupper here.
1745 2018-11-22 Martin Liska <mliska@suse.cz>
1747 * common/config/i386/i386-common.c (processor_names): Add
1748 static assert and add missing "znver2".
1749 (ix86_get_valid_option_values): Add checking assert for null
1750 values and add "native" value if feasible.
1751 * config/i386/i386.h: Do not declare size of processor_names.
1752 * common/config/i386/i386-common.c:
1753 * config/i386/i386.c: Add static assert for size
1754 of processor_cost_table.
1756 2018-11-22 Thomas Preud'homme <thomas.preudhomme@linaro.org>
1758 * target-insns.def (stack_protect_combined_set): Define new standard
1760 (stack_protect_combined_test): Likewise.
1761 * cfgexpand.c (stack_protect_prologue): Try new
1762 stack_protect_combined_set pattern first.
1763 * function.c (stack_protect_epilogue): Try new
1764 stack_protect_combined_test pattern first.
1765 * config/arm/arm.c (require_pic_register): Add pic_reg and compute_now
1766 parameters to control which register to use as PIC register and force
1767 reloading PIC register respectively. Insert in the stream of insns if
1769 (legitimize_pic_address): Expose above new parameters in prototype and
1770 adapt recursive calls accordingly. Use pic_reg if non null instead of
1772 (arm_load_pic_register): Add pic_reg parameter and use it if non null.
1773 (arm_legitimize_address): Adapt to new legitimize_pic_address
1775 (thumb_legitimize_address): Likewise.
1776 (arm_emit_call_insn): Adapt to require_pic_register prototype change.
1777 (arm_expand_prologue): Adapt to arm_load_pic_register prototype change.
1778 (thumb1_expand_prologue): Likewise.
1779 * config/arm/arm-protos.h (legitimize_pic_address): Adapt to prototype
1781 (arm_load_pic_register): Likewise.
1782 * config/arm/predicated.md (guard_addr_operand): New predicate.
1783 (guard_operand): New predicate.
1784 * config/arm/arm.md (movsi expander): Adapt to legitimize_pic_address
1786 (builtin_setjmp_receiver expander): Adapt to thumb1_expand_prologue
1788 (stack_protect_combined_set): New expander..
1789 (stack_protect_combined_set_insn): New insn_and_split pattern.
1790 (stack_protect_set_insn): New insn pattern.
1791 (stack_protect_combined_test): New expander.
1792 (stack_protect_combined_test_insn): New insn_and_split pattern.
1793 (arm_stack_protect_test_insn): New insn pattern.
1794 * config/arm/thumb1.md (thumb1_stack_protect_test_insn): New insn pattern.
1795 * config/arm/unspecs.md (UNSPEC_SP_SET): New unspec.
1796 (UNSPEC_SP_TEST): Likewise.
1797 * doc/md.texi (stack_protect_combined_set): Document new standard
1799 (stack_protect_set): Clarify that the operand for guard's address is
1801 (stack_protect_combined_test): Document new standard pattern name.
1802 (stack_protect_test): Clarify that the operand for guard's address is
1805 2018-11-22 Richard Biener <rguenther@suse.de>
1807 PR tree-optimization/88148
1808 * tree-ssa-loop-niter.c (simplify_replace_tree): Get optional
1809 valueization callback parameter and handle it.
1810 * tree-ssa-loop-niter.h (simplify_replace_tree): Export.
1811 * tree-ssa-sccvn.c (process_bb): Eliminate in loop niter trees.
1813 2018-11-22 Richard Biener <rguenther@suse.de>
1817 * lto-streamer-out.c (lto_is_streamable): Allow CALL_EXPRs
1818 which can appear in size expressions.
1819 * tree-streamer-in.c (unpack_ts_base_value_fields): Stream
1820 CALL_EXPR_BY_DESCRIPTOR.
1821 (streamer_read_tree_bitfields): Stream CALL_EXPR_IFN.
1822 * tree-streamer-out.c (pack_ts_base_value_fields): Stream
1823 CALL_EXPR_BY_DESCRIPTOR.
1824 (streamer_write_tree_bitfields): Stream CALL_EXPR_IFN.
1828 * tree.c (free_lang_data_in_one_sizepos): Free non-gimple-val
1831 2018-11-22 Richard Biener <rguenther@suse.de>
1833 PR tree-optimization/88069
1834 * tree-ssa-sccvn.c (visit_phi): Tweak previous fix to not
1835 apply to default defs.
1837 2018-11-22 Jakub Jelinek <jakub@redhat.com>
1841 * config/i386/i386.c (ix86_option_override_internal): Default
1842 ix86_stack_protector_guard to SSP_TLS only if TARGET_THREAD_SSP_OFFSET
1844 * config/i386/i386.md (stack_protect_set, stack_protect_set_<mode>,
1845 stack_protect_test, stack_protect_test_<mode>): Use empty condition
1846 instead of TARGET_SSP_TLS_GUARD.
1848 2018-11-22 Martin Liska <mliska@suse.cz>
1851 * config/i386/i386.c (ix86_option_override_internal): Error
1852 about usage -mabi=ms and -fsanitize=thread.
1854 2018-11-21 Uros Bizjak <ubizjak@gmail.com>
1857 2013-10-26 Vladimir Makarov <vmakarov@redhat.com>
1860 2013-10-25 Vladimir Makarov <vmakarov@redhat.com>
1862 * lra-spills.c (lra_final_code_change): Remove useless move insns.
1864 2018-11-21 Uros Bizjak <ubizjak@gmail.com>
1867 * function.c (expand_function_end): Do not emit extra blockage insn.
1869 2018-11-21 Lokesh Janghel <lokeshjanghel91@gmail.com>
1872 * config/i386/i386.c (function_value_ms_64): Return AX_REG instead
1873 of FIRST_SSE_REG for 4 or 8 byte modes.
1875 2018-11-21 Jan Hubicka <jh@suse.cz>
1878 * tree.c (fld_decl_context): Break out from ...
1879 (free_lang_data_in_decl): ... here; free TREE_PUBLIC, TREE_PRIVATE
1880 DECL_ARTIFICIAL of TYPE_DECL; do not free TREE_TYPE of TYPE_DECL.
1881 (fld_incomplete_type_of): Build copy of TYP_DECL.
1882 * ipa-devirt.c (free_enum_values): Rename to ...
1883 (free_odr_warning_data): ... this one; free also duplicated TYPE_DECLs
1884 and TREE_TYPEs of TYPE_DECLs.
1885 (get_odr_type): Initialize odr_vtable_hash if needed.
1887 2018-11-21 Alexandre Oliva <oliva@adacore.com>
1889 * final.c (compute_discriminator): Declare. Renamed from...
1890 (maybe_set_discriminator): ... this. Set and return a local.
1891 (override_discriminator): New.
1892 (final_scan_insn_1): Set it.
1893 (notice_source_line): Adjust. Always set discriminator.
1895 2018-11-21 Jakub Jelinek <jakub@redhat.com>
1898 * config/aarch64/atomics.md (@aarch64_compare_and_swap<mode>): Use
1899 rIJ constraint for aarch64_plus_operand rather than rn.
1901 2018-11-21 Renlin Li <renlin.li@arm.com>
1904 * explow.h (get_dynamic_stack_size): Declare it as external.
1905 * explow.c (record_new_stack_level): Remove function static attribute.
1906 * function.c (assign_stack_local_1): Dynamically align the stack slot
1907 addr for parameter copy on the stack.
1909 2018-11-21 Richard Biener <rguenther@suse.de>
1912 * bitmap.c (bitmap_last_set_bit): Refactor to avoid warning.
1913 * Makefile.in (bitmap.o-warn): Remove again.
1915 2018-11-20 Jeff Law <law@redhat.com>
1917 PR tree-optimization/88069
1918 * tree-ssa-dom.c (record_equivalences_from_phis): Propagate away
1919 degenerate virtual PHIs.
1921 2018-11-21 H.J. Lu <hongjiu.lu@intel.com>
1924 * config/i386/sse.md (sse4_1_<code>v8qiv8hi2<mask_name>): Replace
1925 nonimmediate_operand with register_operand.
1926 (avx2_<code>v8qiv8si2<mask_name>): Likewise.
1927 (sse4_1_<code>v4qiv4si2<mask_name>): Likewise.
1928 (sse4_1_<code>v4hiv4si2<mask_name>): Likewise.
1929 (sse4_1_<code>v2qiv2di2<mask_name>): Likewise.
1930 (avx512f_<code>v8qiv8di2<mask_name>): Likewise.
1931 (avx2_<code>v4qiv4di2<mask_name>): Likewise.
1932 (avx2_<code>v4hiv4di2<mask_name>): Likewise.
1933 (sse4_1_<code>v2hiv2di2<mask_name>): Likewise.
1934 (sse4_1_<code>v2siv2di2<mask_name>): Likewise.
1935 (*sse4_1_<code>v8qiv8hi2<mask_name>_1): New pattern.
1936 (*sse4_1_<code>v8qiv8hi2<mask_name>_2): Likewise.
1937 (*avx2_<code>v8qiv8si2<mask_name>_1): Likewise.
1938 (*avx2_<code>v8qiv8si2<mask_name>_2): Likewise.
1939 (*sse4_1_<code>v4qiv4si2<mask_name>_1): Likewise.
1940 (*sse4_1_<code>v4qiv4si2<mask_name>_2): Likewise.
1941 (*sse4_1_<code>v4hiv4si2<mask_name>_1): Likewise.
1942 (*sse4_1_<code>v4hiv4si2<mask_name>_2): Likewise.
1943 (*avx512f_<code>v8qiv8di2<mask_name>_1): Likewise.
1944 (*avx512f_<code>v8qiv8di2<mask_name>_2): Likewise.
1945 (*avx2_<code>v4qiv4di2<mask_name>_1): Likewise.
1946 (*avx2_<code>v4qiv4di2<mask_name>_2): Likewise.
1947 (*avx2_<code>v4hiv4di2<mask_name>_1): Likewise.
1948 (*avx2_<code>v4hiv4di2<mask_name>_2): Likewise.
1949 (*sse4_1_<code>v2hiv2di2<mask_name>_1): Likewise.
1950 (*sse4_1_<code>v2hiv2di2<mask_name>_2): Likewise.
1951 (*sse4_1_<code>v2siv2di2<mask_name>_1): Likewise.
1952 (*sse4_1_<code>v2siv2di2<mask_name>_2): Likewise.
1954 2018-11-21 H.J. Lu <hongjiu.lu@intel.com>
1956 * read-rtl.c (apply_subst_iterator): Handle define_split and
1957 define_insn_and_split.
1959 2018-11-21 Jakub Jelinek <jakub@redhat.com>
1961 PR rtl-optimization/87817
1962 * config/i386/i386.md (bmi2_bzhi_<mode>3, *bmi2_bzhi_<mode>3,
1963 *bmi2_bzhi_<mode>3_1, *bmi2_bzhi_<mode>3_1_ccz): Use IF_THEN_ELSE
1964 in the pattern to avoid triggering UB when operands[2] is zero.
1965 (tbm_bextri_<mode>): New expander. Renamed the old define_insn to ...
1966 (*tbm_bextri_<mode>): ... this.
1968 2018-11-21 Tom de Vries <tdevries@suse.de>
1971 * params.def (HOT_BB_COUNT_FRACTION): Terminate help message with
1974 2018-11-21 Andreas Krebbel <krebbel@linux.ibm.com>
1976 * configure.ac: Add check for Binutils to determine whether vector
1977 load/store alignments hints are being supported.
1978 * config.in: Regenerate.
1979 * configure: Regenerate.
1980 * config/s390/s390.c (print_operand): Support new output
1982 * config/s390/s390.md ("movti"): Append alignment hint output
1983 using the new output modifier 'A'.
1984 * config/s390/vector.md ("mov<mode>", "*vec_tf_to_v1tf")
1985 ("*vec_ti_to_v1ti"): Likewise.
1987 2018-11-20 Martin Sebor <msebor@redhat.com>
1989 * doc/extend.texi (Other Builtins): Add __builtin_has_attribute.
1991 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1994 * ipa-devirt.c (odr_types_equivalent_p): Use operand_equal_p to
1995 compare ENUM values.
1997 2018-11-20 Jakub Jelinek <jakub@redhat.com>
1999 PR tree-optimization/87895
2000 * omp-simd-clone.c (ipa_simd_modify_function_body): When removing
2001 or replacing GIMPLE_RETURN, set EDGE_FALLTHRU on the edge to EXIT.
2002 (simd_clone_adjust): Don't set EDGE_FALLTHRU here. In a loop that
2003 redirects edges to EXIT to edges to incr_bb, iterate while EXIT
2004 has any preds and always use EDGE_PRED (, 0).
2006 2018-11-20 Uros Bizjak <ubizjak@gmail.com>
2009 * mode-switching.c (create_pre_exit): After reload, always split the
2010 fallthrough edge to the exit block.
2012 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
2014 * ipa-devirt.c (add_type_duplicate): Do not ICE on incomplete enums.
2015 * tree.c (build_array_type_1): Forward declare.
2016 (fld_type_variant_equal_p): Add INNER_TYPE parameter.
2017 (fld_type_variant): Likewise.
2018 (fld_simplified_types): New hash.
2019 (fld_process_array_type): New function.
2020 (fld_incomplete_type_of): Handle array and enumeration types.
2021 (fld_simplified_type): Handle simplification of arrays.
2022 (free_lang_data): Allocate and free simplified types hash.
2024 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
2027 * ipa-devirt.c (odr_subtypes_equivalent_p): Report ODR violation
2028 when sybtype already violates ODR.
2029 (get_odr_type): Do not ICE when insert is false and type duplicate
2030 is not registered yet.
2031 (register_odr_type): Be sure to register subtypes first.
2033 2018-11-20 Andreas Krebbel <krebbel@linux.ibm.com>
2035 * config/s390/s390.md ("clztidi2"): Swap the RTX's written to the
2036 DImode parts of the target operand.
2038 2018-11-20 Nathan Sidwell <nathan@acm.org>
2041 * Makefile.in (bitmap.o-warn): Use -Wno-error=array-bounds.
2043 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
2046 * tree.c (free_lang_data_in_cgraph): Add argument fld; break out
2048 (free_lang_data) ... here; update call of free_lang_data_in_cgraph.
2050 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
2053 * ipa-fnsummary.c (pass_ipa_fnsummary): Do not remove functions
2054 * ipa.c (possible_inline_candidate_p): Break out from ..
2055 (process_references): ... here ; drop before_inlining_p;
2056 cleanup handling of alises.
2057 (walk_polymorphic_call_targets): Likewise.
2058 (symbol_table::remove_unreachable_nodes): Likewise.
2059 * passes.c (pass_data_ipa_remove_symbols): New structure.
2060 (pass_ipa_remove_symbols): New pass.
2061 (make_pass_ipa_remove_symbols): New function.
2062 * tree-pass.h (make_pass_ipa_remove_symbols): Declare.
2063 * passes.def (pass_ipa_remove_symbols): Schedule after early passes.
2065 2018-11-20 Richard Biener <rguenther@suse.de>
2067 * tree-vect-stmts.c (vectorizable_condition): Do not get
2068 at else_clause vect def for EXTRACT_LAST_REDUCTION. Remove
2069 pointless vect_is_simple_use calls.
2071 2018-11-20 Richard Biener <rguenther@suse.de>
2073 PR tree-optimization/88074
2074 * tree-ssa-dom.c (pass_dominator::execute): Do not walk
2077 2018-11-20 Richard Biener <rguenther@suse.de>
2079 PR tree-optimization/88069
2080 * tree-ssa-sccvn.c (visit_phi): Do not value-number to unvisited
2081 virtual PHI arguments.
2083 2018-11-20 Ilya Leoshkevich <iii@linux.ibm.com>
2086 * config/s390/s390.md: Skip LT(G) peephole when literal pool is
2088 * rtl.h (contains_constant_pool_address_p): New function.
2089 * rtlanal.c (contains_constant_pool_address_p): Likewise.
2091 2018-11-20 Richard Biener <rguenther@suse.de>
2094 * alias.c (component_uses_parent_alias_set_from): Remove
2095 alias-set zero and TYPE_TYPELESS_STORAGE case both already
2096 handled in other ways.
2098 2018-11-20 Richard Biener <rguenther@suse.de>
2100 PR tree-optimization/88087
2101 * tree-ssa-pre.c (create_expression_by_pieces): Re-materialize
2103 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Remember
2106 2018-11-20 Richard Biener <rguenther@suse.de>
2109 * tree-data-ref.c (lambda_matrix_right_hermite): Use abs_hwi.
2111 2018-11-19 Jan Hubicka <hubicka@ucw.cz>
2114 * ipa-devirt.c (free_enum_values): Do not ICE on ODR vilations.
2116 2018-11-19 Sandra Loosemore <sandra@codesourcery.com>
2119 * doc/invoke.texi (Link Options): Mention shared libraries
2120 in documentation for the -l option. Simplify discussion and
2121 point to the system linker documentation for details.
2123 2018-11-19 Peter Bergner <bergner@linux.ibm.com>
2125 PR rtl-optimization/88033
2126 * ira-lives.c (non_conflicting_reg_copy_p): Skip copies from a register
2127 to itself. Use HARD_REGISTER_NUM_P.
2129 2018-11-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2131 * tree-vect-loop.c (vect_transform_loop): Disable further unrolling
2132 of the loop if vf is non-constant.
2134 2018-11-19 David Malcolm <dmalcolm@redhat.com>
2136 PR tree-optimization/87025
2137 * dumpfile.c (dump_context::begin_scope): Call end_any_optinfo
2138 immediately after creating the scope optinfo.
2139 (selftest::test_pr87025): New function.
2140 (selftest::dumpfile_c_tests): Call it.
2141 * optinfo-emit-json.cc (optrecord_json_writer::pop_scope): Assert
2142 that we're not popping the top-level records array.
2143 * optinfo.cc (optinfo::handle_dump_file_kind): Assert that we're
2144 not changing the kind of a "scope" optinfo.
2146 2018-11-19 David Malcolm <dmalcolm@redhat.com>
2148 PR tree-optimization/87025
2149 * dump-context.h: Include "optinfo.h".
2150 (class optrecord_json_writer): New forward decl.
2151 (dump_context::forcibly_enable_optinfo_p): Delete.
2152 (dump_context::optinfo_enabled_p): New member function.
2153 (dump_context::optimization_records_enabled_p): New member
2155 (dump_context::set_json_writer): New member function.
2156 (dump_context::emit_optinfo): New member function.
2157 (dump_context::m_forcibly_enable_optinfo): Delete.
2158 (dump_context::m_json_writer): New member data.
2159 * dumpfile.c (dump_context::set_json_writer): New member function.
2160 (dump_context::finish_any_json_writer): New member function.
2161 (dump_context::end_scope): Replace call to
2162 optimization_records_maybe_pop_dump_scope with call to
2163 m_json_writer->pop_scope.
2164 (dump_context::optinfo_enabled_p): New member function.
2165 (dump_context::end_any_optinfo): Replace call to optinfo::emit with call
2166 to dump_context::emit_optinfo.
2167 (dump_context::emit_optinfo): New member function.
2168 (temp_dump_context::temp_dump_context): Replace
2169 m_forcibly_enable_optinfo with call to set_json_writer.
2170 (temp_dump_context::~temp_dump_context): Clean up any json writer.
2171 * optinfo-emit-json.cc (class optrecord_json_writer): Move to
2173 (the_json_writer): Delete.
2174 (optimization_records_start): Delete.
2175 (optimization_records_finish): Delete.
2176 (optimization_records_enabled_p): Delete, in favor of
2177 dump_context::optimization_records_enabled_p.
2178 (optimization_records_maybe_record_optinfo): Delete.
2179 (optimization_records_maybe_pop_dump_scope): Delete.
2180 * optinfo-emit-json.h: Include "json.h". Delete forward
2182 (optimization_records_start): Delete.
2183 (optimization_records_finish): Delete.
2184 (optimization_records_enabled_p): Delete.
2185 (optimization_records_maybe_record_optinfo): Delete.
2186 (optimization_records_maybe_pop_dump_scope): Delete.
2187 (class optrecord_json_writer): Move here from
2188 optinfo-emit-json.cc.
2189 * optinfo.cc (optinfo::emit_for_opt_problem): Replace call
2190 to optinfo::emit with call to dump_context::emit_optinfo.
2191 (optinfo::emit): Delete, in favor of dump_context::emit_optinfo.
2192 (optinfo_enabled_p): Delete, in favor of
2193 dump_context::optinfo_enabled_p.
2194 (optinfo_wants_inlining_info_p): Update for conversion o
2195 optimization_records_enabled_p to a member function of
2197 * optinfo.h (optinfo_enabled_p): Delete, in favor of
2198 dump_context::optinfo_enabled_p.
2199 (optinfo::emit): Delete, in favor of dump_context::emit_optinfo.
2200 * toplev.c: Include "dump-context.h".
2201 (compile_file): Replace call to optimization_records_finish with
2202 dump_context::finish_any_json_writer.
2203 (do_compile): Replace call to optimization_records_start with
2204 conditionally creating a optrecord_json_writer for the
2207 2018-11-19 Jakub Jelinek <jakub@redhat.com>
2209 PR tree-optimization/88071
2210 * tree-vect-loop.c (vectorize_fold_left_reduction): Pass true instead
2211 of false as last argument to gsi_remove.
2212 * tree-vect-stmts.c (vect_finish_replace_stmt): Pass true instead of
2213 false as last argument to gsi_replace.
2216 * omp-expand.c: Don't include debug.h.
2217 (adjust_context_and_scope): Add REGION argument. Find DECL_CONTEXT
2218 from innermost outer parallel, task, teams or target that has a
2219 child_fn set, or, if there is no such outer region, use
2220 current_function_decl. Do the DECL_CONTEXT adjustment regardless of
2221 whether a suitable BLOCK is found or not.
2222 (expand_parallel_call, expand_teams_call): Don't call
2223 adjust_context_and_scope here.
2224 (grid_expand_target_grid_body): Revert 2017-01-25 changes.
2225 (expand_omp_taskreg, expand_omp_target): Likewise. Call
2226 adjust_context_and_scope.
2227 * dwarf2out.c (dwarf2out_early_global_decl): For
2228 decl_function_context recurse instead of calling dwarf2out_decl.
2230 2018-11-20 Eric Botcazou <ebotcazou@adacore.com>
2232 PR rtl-optimization/85925
2233 * rtl.h (word_register_operation_p): New predicate.
2234 * combine.c (record_dead_and_set_regs_1): Only apply specific handling
2235 for WORD_REGISTER_OPERATIONS targets to word_register_operation_p RTX.
2236 * rtlanal.c (nonzero_bits1): Likewise. Adjust couple of comments.
2237 (num_sign_bit_copies1): Likewise.
2239 2018-11-19 Richard Biener <rguenther@suse.de>
2242 * tree.c (free_lang_data_in_one_sizepos): Free non-gimple-val
2245 2018-11-19 Eric Botcazou <ebotcazou@adacore.com>
2247 * compare-elim.c (struct comparison): Add not_in_a field.
2248 (is_not): New static function.
2249 (strip_not): Likewise.
2250 (conforming_compare): Handle a NOT in the first operand.
2251 (can_eliminate_compare): Likewise.
2252 (find_comparison_dom_walker::before_dom_children): Likewise.
2253 (try_eliminate_compare): Likewise.
2254 * config/visium/visium.md (negsi2_insn_set_carry): Turn into...
2255 (neg<mode>2_insn_set_carry): ...this and add missing NEG operation.
2257 2018-11-19 Jonathan Wakely <jwakely@redhat.com>
2259 * doc/extend.texi (Common Type Attributes): Fix typos.
2261 2018-11-19 Martin Liska <mliska@suse.cz>
2263 PR gcov-profile/88045
2264 * coverage.c (coverage_begin_function): Add assert.
2266 2018-11-18 Sandra Loosemore <sandra@codesourcery.com>
2269 * doc/extend.texi (Common Function Attributes): Document that
2270 no_instrument_function applies to -p and -pg, too.
2271 * doc/invoke.texi (Instrumentation Options): Add cross-references
2272 to docs for -p, -pg, and -finstrument-functions.
2274 2018-11-18 Alan Modra <amodra@gmail.com>
2276 * config/gnu-user.h (GNU_USER_TARGET_CRTI): Define.
2277 (GNU_USER_TARGET_STARTFILE_SPEC): Use it here.
2278 (GNU_USER_TARGET_CRTN): Define.
2279 (GNU_USER_TARGET_ENDFILE_SPEC): Use it here.
2280 (GNU_USER_TARGET_LINK_GCC_C_SEQUENCE_SPEC): Define.
2282 2018-11-18 Alan Modra <amodra@gmail.com>
2284 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Delete
2285 !HAVE_LD_PIE variant.
2286 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
2288 2018-11-17 Nick Clifton <nickc@redhat.com>
2289 Sandra Loosemore <sandra@codesourcery.com>
2292 * doc/invoke.texi (Overall Options): Document that --help and
2293 --help=<value> options cannot be combined.
2295 2018-11-17 Jim Wilson <jimw@sifive.com>
2297 * config/riscv/riscv.c (epilogue_cfa_sp_offset): New.
2298 (riscv_restore_reg): If restoring HARD_FRAME_POINTER_REGNUM, and
2299 epilogue_cfa_sp_offset set, then add REG_CFA_DEF_CFA regnote.
2300 (riscv_expand_epilogue): Initialize epilogue_cfa_sp_offset. Set it
2301 to step2 if frame_pointer_needed and step1 is 0.
2303 2018-11-17 Sandra Loosemore <sandra@codesourcery.com>
2306 * doc/cpp.texi (System Headers): Add note about implicit
2307 extern "C" block on targets that define SYSTEM_IMPLICIT_EXTERN_C.
2309 2018-11-17 Sandra Loosemore <sandra@codesourcery.com>
2312 * doc/invoke.texi (C++ Dialect Options): Clarify usage of
2313 -fno-implicit-templates.
2315 2018-11-17 Jakub Jelinek <jakub@redhat.com>
2317 PR tree-optimization/87546
2318 * tree-vect-patterns.c (vect_look_through_possible_promotion): Add
2319 min_precision variable, initially set it to orig_precision, only does
2320 something if op_type's precision is <= min_precision and update
2321 min_precision whenever calling set_op.
2323 2018-11-16 Jan Hubicka <hubicka@ucw.cz>
2326 * ipa-devirt.c (warn_odr): Look for main variant to get TYPE_DECL.
2328 2018-11-16 Sandra Loosemore <sandra@codesourcery.com>
2330 * doc/invoke.texi (Option Summary): Fix whitespace and line
2331 breaks in @gccoptlist environments.
2332 (Warning Options): Likewise.
2333 (Optimize Options): Likewise.
2334 (PowerPC SPE Options): Likewise.
2335 (RS/6000 and PowerPC Options): Likewise.
2337 2018-11-16 Jeff Law <law@redhat.com>
2339 * config/mn10300/mn10300.md (adddi3_degenerate): Remove bogus
2342 2018-11-16 Sandra Loosemore <sandra@codesourcery.com>
2345 * doc/invoke.texi (Optimize Options): Update options enabled by
2346 fprofile-generate, -fprofile-use, and -fauto-profile.
2348 2018-11-16 Jozef Lawrynowicz <jozef.l@mittosystems.com>
2351 * target-def.h: Initialize TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP.
2352 Add them to the TARGET_ASM_{,UN}ALIGNED_INT_OP structs.
2353 * target.def: Enumerate TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP in
2355 * target.h: Add psi, pdi, pti to struct asm_int_op definition.
2356 * targhooks.c (default_print_patchable_function_entry): Assert
2357 asm_int_op does not return a NULL string.
2358 * varasm.c (integer_asm_op): Return the op for a partial int type
2359 when the requested size does not correspond to an integer type.
2360 * config/msp430/msp430.c: Initialize TARGET_ASM_{,UN}ALIGNED_PSI_OP.
2361 * doc/tm.texi: Regenerate.
2363 2018-11-12 Jason Merrill <jason@redhat.com>
2365 * gimplify.c (gimplify_case_label_expr): Handle hot/cold attributes.
2367 2018-11-16 Michael Meissner <meissner@linux.ibm.com>
2369 * config/rs6000/constraints.md (wF constraint): Remove power9
2370 fusion documentation. Just document wF's use for power8 fusion.
2371 * config/rs6000/predicates.md (p9_fusion_reg_operand): Delete.
2372 (fusion_gpr_addis): Delete power9 fusion support. Change power8
2373 fusion support to require the upper 12 bits to be all 0's or all
2375 (fusion_gpr_mem_load): Add comment.
2376 (fusion_addis_mem_combo_load): Remove power9 fusion support.
2377 (fusion_addis_mem_combo_store): Delete.
2378 (fusion_offsettable_mem_operand): Delete.
2379 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Do not set
2381 (ISA_3_0_MASKS_SERVER): Delete power9 fusion.
2382 (POWERPC_MASKS): Delete power9 fusion.
2383 * config/rs6000/rs6000-protos.h (emit_fusion_load_store): Delete.
2384 (fusion_p9_p): Delete.
2385 (expand_fusion_p9_load): Delete.
2386 (expand_fusion_p9_store): Delete.
2387 (emit_fusion_p9_load): Delete.
2388 (emit_fusion_p9_store): Delete.
2389 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Delete power9
2391 (rs6000_option_override_internal): Set power8 fusion based on
2392 whether we are tuning for power8. Delete power9 fusion support.
2393 (rs6000_opt_masks): Delete -mpower9-fusion switch.
2394 (emit_fusion_load): Rename emit_fusion_load_store to
2395 emit_fusion_load, and drop fusion store support. Update callers.
2396 (emit_fusion_load_store): Likewise.
2397 (emit_fusion_gpr_load): Likewise.
2398 (fusion_p9_p): Delete.
2399 (expand_fusion_p9_load): Delete.
2400 (expand_fusion_p9_store): Delete.
2401 (emit_fusion_p9_load): Delete.
2402 (emit_fusion_p9_store): Delete.
2403 * config/rs6000/rs6000.md (UNSPEC_FUSION_P9): Delete.
2404 (GPR_FUSION): Delete.
2405 (FPR_FUSION): Delete.
2406 (power9 fusion peephole2s): Delete.
2407 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Delete.
2408 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Delete.
2409 (fusion_vsx_<P:mode>_<FPR_FUSION:mode>_load): Delete.
2410 (fusion_vsx_<P:mode>_<FPR_FUSION:mode>_store): Delete.
2411 (fusion_p9_<mode>_constant): Delete.
2412 * config/rs6000/rs6000.opt (-mpower9-fusion): Delete undocumented
2413 power9 fusion switch.
2414 * doc/md.texi (PowerPC constraints): Update wF constraint
2415 documentation for power8 fusion only.
2417 2018-11-16 Jakub Jelinek <jakub@redhat.com>
2419 PR rtl-optimization/87475
2420 * cfgrtl.c (patch_jump_insn): Allow redirection failure for
2421 CROSSING_JUMP_P insns.
2422 (cfg_layout_redirect_edge_and_branch): Don't ICE if ret is NULL.
2424 2018-11-16 Uros Bizjak <ubizjak@gmail.com>
2427 * config/i386/i386.md (floatunsdidf2): Allow only 64bit AVX512F targets.
2428 * config/i386/sse.md (UNSPEC_MOVDI_TO_SSE): New UNSPEC.
2429 (movdi_to_sse): Rewrite using UNSPEC_MOVDI_TO_SSE unspec.
2431 2018-11-16 Jakub Jelinek <jakub@redhat.com>
2434 * optabs.c (expand_binop): For op0_mode use GET_MODE (op0), unless it
2435 is VOIDmode, in which case use int_mode. Similarly for op1_mode.
2437 2018-11-16 Eric Botcazou <ebotcazou@adacore.com>
2439 * tree-switch-conversion.h (switch_decision_tree::emit_case_nodes): Add
2440 location_t parameter.
2441 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
2442 (switch_decision_tree::do_jump_if_equal): Likewise.
2443 * tree-switch-conversion.c (switch_decision_tree::emit): Pass location
2444 of switch statement to emit_case_nodes.
2445 (switch_decision_tree::emit_cmp_and_jump_insns): Add LOC parameter and
2446 set it on the newly built GIMPLE comparison statement.
2447 (switch_decision_tree::do_jump_if_equal): Likewise.
2448 (switch_decision_tree::emit_case_nodes): Add LOC parameter and pass it
2449 in calls to do_jump_if_equal as well as recursive calls.
2451 2018-11-16 Andi Kleen <ak@linux.intel.com>
2453 * config/i386/i386.md: Allow memory operands to ptwrite.
2455 2018-11-16 Michael Matz <matz@suse.de>
2457 * system.h (PRsa): New macro.
2458 (SIZE_AMOUNT): Cast number to uint64_t.
2459 * alloc-pool.h (pool_usage::dump): Don't use %zu but PRsa.
2460 (pool_usage::dump_footer): Likewise and also use PRIu64.
2461 * bitmap.h (bitmap_usage::dump): Likewise.
2462 * ggc-common.c (ggc_usage::dump): Likewise.
2463 * ggc-page.c (ggc_print_statistics): Likewise.
2464 * input.c (dump_line_table_statistics): Likewise.
2465 * mem-stats.h (mem_usage::dump): Likewise.
2466 (mem_usage::dump_footer): Likewise.
2467 * rtl.c (dump_rtx_statistics): Likewise.
2468 * tree-cfg.c (dump_cfg_stats): Likewise.
2469 * tree-dfa.c (dump_dfa_stats): Likewise.
2470 * tree-phinodes.c (phinodes_print_statistics): Likewise.
2471 * tree-ssanames (ssanames_print_statistics): Likewise.
2472 * vec.c (vec_usage::dump): Likewise.
2473 (vec_usage::dump_footer): Likewise.
2475 2018-11-16 Richard Biener <rguenther@suse.de>
2477 PR tree-optimization/88011
2478 * tree-vrp.c (extract_range_from_binary_expr): Fix error in
2479 replacing set_value_range_to_undefined and
2480 set_value_range_to_varying with method calls.
2482 2018-11-16 Ilya Leoshkevich <iii@linux.ibm.com>
2484 * config/s390/s390.md
2485 (*r<noxa>sbg_<mode>_srl_bitmask): Do not delegate arithmetic to
2487 (*r<noxa>sbg_<mode>_sll): Likewise.
2488 (*r<noxa>sbg_<mode>_srl): Likewise.
2489 (*r<noxa>sbg_sidi_srl): New pattern.
2491 2018-11-16 Jerome Lambourg <lambourg@adacore.com>
2493 * config/vxworks.h (TARGET_VXWORKS_HAVE_CTORS_DTORS): New macro.
2494 Default to TARGET_VXWORKS_RTP.
2495 (SUPPORTS_INIT_PRIORITY): Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
2496 of TARGET_VXWORKS_RTP.
2497 * config/vxworks.c: Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
2498 of TARGET_VXWORKS_RTP to set targetm.have_ctors_dtors.
2500 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
2503 * doc/extend.texi (Common Type Attributes): Make it explicit
2504 that attribute "packed" can apply to C++ classes.
2506 2018-11-15 Martin Sebor <msebor@redhat.com>
2510 * tree.c (type_argument_type): New function.
2511 * tree.h (type_argument_type): Declare it.
2512 * gcc/doc/extend.texi (alloc_align): Update and clarify.
2513 (alloc_size, nonnull, sentinel): Same.
2515 2018-11-15 Andrew Stubbs <ams@codesourcery.com>
2516 Kwok Cheung Yeung <kcy@codesourcery.com>
2518 * tree-vect-stmts.c (vectorizable_store): Don't ICE when
2519 int_mode_for_size fails.
2520 (vectorizable_load): Likewise.
2522 2018-11-15 David Malcolm <dmalcolm@redhat.com>
2524 * doc/ux.texi (Group logically-related diagnostics): Move
2525 discussion of auto_diagnostic_group into this new subsection.
2526 Give an example of where this grouping is used.
2528 2018-11-15 David Malcolm <dmalcolm@redhat.com>
2531 * Makefile.in (OBJS): Move json.o to...
2532 (OBJS-libcommon): ...here and add diagnostic-format-json.o.
2533 * common.opt (fdiagnostics-format=): New option.
2534 (diagnostics_output_format): New enum.
2535 * diagnostic-format-json.cc: New file.
2536 * diagnostic.c (default_diagnostic_final_cb): New function, taken
2537 from start of diagnostic_finish.
2538 (diagnostic_initialize): Initialize final_cb to
2539 default_diagnostic_final_cb.
2540 (diagnostic_finish): Move "being treated as errors" messages to
2541 default_diagnostic_final_cb. Call any final_cb.
2542 (default_diagnostic_finalizer): Add diagnostic_t param.
2543 (diagnostic_report_diagnostic): Pass "orig_diag_kind" to
2544 diagnostic_finalizer callback.
2545 * diagnostic.h (enum diagnostics_output_format): New enum.
2546 (diagnostic_finalizer_fn): Reimplement, adding diagnostic_t param.
2547 (struct diagnostic_context): Add "final_cb".
2548 (default_diagnostic_finalizer): Add diagnostic_t param.
2549 (diagnostic_output_format_init): New decl.
2550 * doc/invoke.texi (-fdiagnostics-format): New option.
2551 * dwarf2out.c (gen_producer_string): Ignore
2552 OPT_fdiagnostics_format_.
2553 * gcc.c (driver_handle_option): Handle OPT_fdiagnostics_format_.
2554 * lto-wrapper.c (append_diag_options): Ignore it.
2555 * opts.c (common_handle_option): Handle it.
2557 2018-11-15 David Malcolm <dmalcolm@redhat.com>
2559 PR tree-optimization/88015
2560 * graphite-isl-ast-to-gimple.c
2561 (translate_isl_ast_to_gimple::scop_to_isl_ast): Add missing check
2563 * graphite-sese-to-poly.c (build_poly_scop): Likewise.
2565 2018-11-15 Richard Biener <rguenther@suse.de>
2568 * gimple.c (gimple_call_flags): Union flags from decl, type
2570 * trans-mem.c (is_tm_pure_call): Simplify.
2572 2018-11-15 Richard Biener <rguenther@suse.de>
2574 PR tree-optimization/88031
2575 * tree-vect-loop.c (vectorizable_reduction): Move check
2576 for multiple types earlier so we get the expected dump.
2577 Simplify calls to vectorizable_condition.
2578 * tree-vect-stmts.h (vectorizable_condition): Update prototype.
2579 * tree-vect-stmts.c (vectorizable_condition): Instead of
2580 reduc_def and reduc_index take just a flag. Simplify
2581 code-generation now that we can rely on the defs being set up.
2582 (vectorizable_comparison): Remove unused argument.
2584 2018-11-15 Richard Biener <rguenther@suse.de>
2586 PR tree-optimization/88030
2587 * tree-complex.c (need_eh_cleanup): New global.
2588 (update_complex_assignment): Mark blocks that need EH update.
2589 (expand_complex_comparison): Likewise.
2590 (tree_lower_complex): Allocate and deallocate need_eh_cleanup,
2591 perform EH cleanup and schedule CFG cleanup if that did anything.
2593 2018-11-15 Jakub Jelinek <jakub@redhat.com>
2595 PR rtl-optimization/88018
2596 * cfgrtl.c (fixup_abnormal_edges): Guard moving insns to fallthru edge
2597 on the presence of fallthru edge, rather than if it is a USE or not.
2599 2018-11-15 Richard Biener <rguenther@suse.de>
2602 * tree-data-ref.c (analyze_miv_subscript): Guard calls to
2603 analyze_subscript_affine_affine properly.
2605 2018-11-15 Xianmiao Qu <xianmiao_qu@c-sky.com>
2607 * config/csky/csky-linux-elf.h (LINUX_DYNAMIC_LINKER): Remove.
2608 (GLIBC_DYNAMIC_LINKER): Define.
2609 (LINUX_TARGET_LINK_SPEC): Update the dynamic linker's name.
2611 2018-11-15 Bin Cheng <bin.cheng@linux.alibaba.com>
2613 PR tree-optimization/84648
2614 * tree-ssa-loop-niter.c (adjust_cond_for_loop_until_wrap): New.
2615 (number_of_iterations_cond): Adjust exit cond for loop-until-wrap case
2616 by calling adjust_cond_for_loop_until_wrap.
2618 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
2621 * doc/extend.texi (Common Function Attributes): Clarify linker
2622 restrictions on "aligned" attribute.
2623 (Common Variable Attributes): Likewise. Mention that linker
2624 restrictions don't apply to stack-allocated variables.
2626 2018-11-15 Joern Rennecke <joern.rennecke@riscy-ip.com>
2628 * gensupport.c (add_predicate_code): Properly handle ZERO_EXTRACT
2631 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
2635 * doc/invoke.texi (Optimize Options): Remove bad example about
2636 interaction between -flto and -O. Replace it with a note that
2637 you need to compile with -O and not just link. Copy-edit -flto
2638 discussion to reduce verbiage and improve flow.
2640 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
2643 * doc/invoke.texi (Optimize Options): Clarify that -O0 and -Og
2644 also suppress many optimizations. Alphabetize option lists for
2645 -O1, -O2, and -Os. Add list of options disabled with -Og, and
2646 correct documentation for those options to say that.
2647 * opts.c (default_options_table): Sort table by level and option
2648 name, to make it easier to correlate to the manual.
2650 2018-11-14 Uros Bizjak <ubizjak@gmail.com>
2652 * config/i386/i386.c (ix86_print_operand_address_as): Simplify
2653 printing of the names of segment registers.
2655 2018-11-14 Nathan Sidwell <nathan@acm.org>
2659 * dwarf2out.c (dwarf2out_finish): Apply resolve_addr to comdat
2662 2018-11-14 David Malcolm <dmalcolm@redhat.com>
2664 * Makefile.in (CFLAGS-optinfo-emit-json.o): Add $(ZLIBINC).
2666 2018-11-14 Jakub Jelinek <jakub@redhat.com>
2668 P1236R1 - Signed integers are two's complement
2669 * doc/invoke.texi (Wshift-overflow): Adjust documentation for c++2a.
2672 * hash-map.h (hash_map::iterator::reference_pair): New class.
2673 (hash_map::iterator::operator*): Return it rather than std::pair.
2675 2018-11-14 Jeff Law <law@redhat.com>
2677 * optabs.c (expand_binop): Pass INT_MODE to operand_subword_force
2678 iff the operand is a constant.
2680 2018-11-14 Aldy Hernandez <aldyh@redhat.com>
2682 * gimple-ssa-evrp-analyze.c
2683 (evrp_range_analyzer::record_ranges_from_incoming_edge): Rename
2684 ignore_equivs_equal_p to equal_p.
2685 * ipa-cp.c (meet_with_1): Use equal_p instead of
2686 ignore_equivs_equal_p.
2687 * ipa-prop.c (ipa_vr_ggc_hash_traits::equal): Same.
2688 * tree-vrp.c (value_range::ignore_equivs_equal_p): Remove.
2689 (value_range::operator==): Remove.
2690 (value_range::operator!=): Remove.
2691 (vrp_prop::visit_stmt): Use equal_p.
2692 * tree-vrp.h (value_range): Remove operator==, operator!=,
2693 ignore_equivs_equal_p.
2694 * vr-values.c (update_value_range): Use equal_p.
2696 2018-11-14 Michael Matz <matz@suse.de>
2699 * gimplify.c (collect_fallthrough_labels): Add new argument,
2700 return location via that, don't modify statements.
2701 (warn_implicit_fallthrough_r): Adjust call, don't use
2702 statement location directly.
2704 2018-11-14 Richard Biener <rguenther@suse.de>
2707 * tree-data-ref.c (split_constant_offset): Add wrapper
2708 allocating a cache hash-map.
2709 (split_constant_offset_1): Cache results of expanding
2710 expressions from SSA def stmts.
2712 2018-11-14 Richard Biener <rguenther@suse.de>
2715 * tree-data-ref.c (lambda_matrix_row_add): Change const1 argument
2717 (lambda_vector_mult_const): Likewise.
2718 (lambda_matrix_right_hermite): Use lambda_int temporaries.
2720 2018-11-14 Wilco Dijkstra <wdijkstr@arm.com>
2721 Jackson Woodruff <jackson.woodruff@arm.com>
2723 PR 71026/tree-optimization
2724 * match.pd: Simplify floating point comparisons.
2726 2018-11-14 Jakub Jelinek <jakub@redhat.com>
2728 PR rtl-optimization/87817
2729 * config/i386/i386.c (ix86_fold_builtin): For _bzhi_u{32,64} if
2730 last argument has low 8 bits clear, fold to 0.
2732 2018-11-14 Iain Sandoe <iain@sandoe.co.uk>
2734 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p): Cast
2735 MAX_OFILE_ALIGNMENT as needed.
2737 2018-11-14 Richard Biener <rguenther@suse.de>
2739 PR tree-optimization/88019
2740 * graphite-isl-ast-to-gimple.c (ternary_op_to_tree): Rewrite
2741 COND_EXPR arguments to non-trapping overflow.
2743 2018-11-14 Claudiu Zissulescu <claziss@synopsys.com>
2745 * config/arc/arc.h (reg_class): Reorder registers classes, remove
2746 unused register classes.
2747 (REG_CLASS_NAMES): Likewise.
2748 (REG_CLASS_CONTENTS): Likewise.
2749 (FIXED_REGISTERS): Make lp_count fixed.
2750 (BASE_REG_CLASS): Remove ACC16_BASE_REGS reference.
2751 (PROGRAM_COUNTER_REGNO): Remove.
2752 * config/arc/arc.c (arc_conditional_register_usage): Remove unused
2753 register classes, use constants for register numbers, remove
2754 reg_class_contents references.
2755 (arc_process_double_reg_moves): Add asserts.
2756 (arc_secondary_reload): Remove LPCOUNT_REG reference, use
2757 lra_in_progress predicate.
2758 (arc_init_reg_tables): Remove unused register classes.
2759 (arc_register_move_cost): Likewise.
2760 (arc_preferred_reload_class): Likewise.
2761 (hwloop_optimize): Update rtx patterns involving lp_count register.
2762 (arc_return_address_register): Rename ILINK1, INLINK2 regnums macros.
2763 * config/arc/constraints.md ("c"): Choose between GENERAL_REGS and
2764 CHEAP_CORE_REGS. Former one will be used for LRA.
2765 ("Rac"): Choose between GENERAL_REGS and ALL_CORE_REGS. Former
2766 one will be used for LRA.
2767 ("w"): Choose between GENERAL_REGS and WRITABLE_CORE_REGS. Former
2768 one will be used for LRA.
2769 ("W"): Choose between GENERAL_REGS and MPY_WRITABLE_CORE_REGS.
2770 Former one will be used for LRA.
2771 ("f"): Delete constraint.
2774 ("l"): Change it from register constraint to constraint.
2775 * config/arc/arc.md (movqi_insn): Remove unsed lp_count constraints.
2776 (movhi_insn): Likewise.
2777 (movsi_insn): Update pattern.
2780 (stack_tie): Remove 'b' constraint letter.
2782 (R9_REG, R15_REG, R16_REG, R25_REG): Likewise.
2783 (R32_REG, R40_REG, R41_REG, R42_REG, R43_REG, R44_REG): Likewise.
2784 (R57_REG, R59_REG, PCL_REG): Likewise.
2785 (ILINK1_REGNUM): Renamed to ILINK1_REG.
2786 (ILINK2_REGNUM): Renamed to ILINK2_REG.
2788 (SP_REGS): Likewise.
2789 (Rcw): Remove unused reg classes.
2790 * config/arc/predicates.md (dest_reg_operand): Just default on
2791 register_operand predicate.
2792 (mpy_dest_reg_operand): Likewise.
2793 (move_dest_operand): Use macros instead of constants.
2794 * config/arc/arc.opt (mlra): Switch to lra as default.
2796 2018-11-14 Richard Biener <rguenther@suse.de>
2798 PR tree-optimization/87974
2799 * tree-vect-loop.c (vectorizable_reduction): When computing
2800 the vectorized reduction PHI vector type ignore constant
2803 2018-11-14 Jakub Jelinek <jakub@redhat.com>
2805 PR tree-optimization/87977
2806 * tree-ssa-math-opts.c (optimize_recip_sqrt): Don't reuse division
2807 stmt, build a new one and replace the old one with it. Formatting fix.
2808 Call release_ssa_name (x) if !has_other_use and !delete_div.
2809 (pass_cse_reciprocals::execute): Before calling optimize_recip_sqrt
2810 verify lhs of stmt is still def.
2812 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
2814 PR rtl-optimization/87507
2815 * lower-subreg.c (operand_for_swap_move_operator): New function.
2816 (simple_move): Strip simple operators.
2817 (find_pseudo_copy): Likewise.
2818 (resolve_operand_for_swap_move_operator): New function.
2819 (resolve_simple_move): Strip simple operators and swap operands.
2821 2018-11-13 Jakub Jelinek <jakub@redhat.com>
2823 PR tree-optimization/87898
2824 * omp-simd-clone.c (ipa_simd_modify_stmt_ops): Formatting fix.
2825 (ipa_simd_modify_function_body): Remove debug stmts where the first
2826 argument was changed into a non-decl.
2828 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
2830 PR rtl-optimization/87899
2831 * lra-lives.c (start_living): Update white space in comment.
2832 (enum point_type): New.
2833 (sparseset_contains_pseudos_p): New function.
2834 (update_pseudo_point): Likewise.
2835 (make_hard_regno_live): Use HARD_REGISTER_NUM_P macro.
2836 (make_hard_regno_dead): Likewise. Remove ignore_reg_for_conflicts
2837 handling. Move early exit after adding conflicts.
2838 (mark_pseudo_live): Use HARD_REGISTER_NUM_P macro. Add early exit
2839 if regno is already live. Remove all handling of program points.
2840 (mark_pseudo_dead): Use HARD_REGISTER_NUM_P macro. Add early exit
2841 after adding conflicts. Remove all handling of program points and
2842 ignore_reg_for_conflicts.
2843 (mark_regno_live): Use HARD_REGISTER_NUM_P macro. Remove return value
2844 and do not guard call to mark_pseudo_live.
2845 (mark_regno_dead): Use HARD_REGISTER_NUM_P macro. Remove return value
2846 and do not guard call to mark_pseudo_dead.
2847 (check_pseudos_live_through_calls): Use HARD_REGISTER_NUM_P macro.
2848 (process_bb_lives): Use HARD_REGISTER_NUM_P and HARD_REGISTER_P macros.
2849 Use new function update_pseudo_point. Handle register copies by
2850 removing the source register from the live set. Handle INOUT operands.
2851 Update to the next program point using the unused_set, dead_set and
2853 (lra_create_live_ranges_1): Use HARD_REGISTER_NUM_P macro.
2855 2018-11-13 David Malcolm <dmalcolm@redhat.com>
2857 * builtins.c: Replace "source_location" with "location_t".
2858 * diagnostic-show-locus.c: Likewise.
2859 * diagnostic.c: Likewise.
2860 * dumpfile.c: Likewise.
2861 * gcc-rich-location.h: Likewise.
2862 * genmatch.c: Likewise.
2863 * gimple.h: Likewise.
2864 * gimplify.c: Likewise.
2865 * input.c: Likewise.
2866 * input.h: Likewise. Eliminate the typedef.
2867 * omp-expand.c: Likewise.
2868 * selftest.h: Likewise.
2869 * substring-locations.h (get_source_location_for_substring):
2871 (get_location_within_string): ...this.
2872 * tree-cfg.c: Replace "source_location" with "location_t".
2873 * tree-cfgcleanup.c: Likewise.
2874 * tree-diagnostic.c: Likewise.
2875 * tree-into-ssa.c: Likewise.
2876 * tree-outof-ssa.c: Likewise.
2877 * tree-parloops.c: Likewise.
2878 * tree-phinodes.c: Likewise.
2879 * tree-phinodes.h: Likewise.
2880 * tree-ssa-loop-ivopts.c: Likewise.
2881 * tree-ssa-loop-manip.c: Likewise.
2882 * tree-ssa-phiopt.c: Likewise.
2883 * tree-ssa-phiprop.c: Likewise.
2884 * tree-ssa-threadupdate.c: Likewise.
2885 * tree-ssa.c: Likewise.
2886 * tree-ssa.h: Likewise.
2887 * tree-vect-loop-manip.c: Likewise.
2889 2018-11-13 Michael Ploujnikov <michael.ploujnikov@oracle.com>
2891 * doc/extend.texi: Fix typo in the weakref description.
2893 2018-11-13 Richard Biener <rguenther@suse.de>
2895 PR tree-optimization/86991
2896 * tree-vect-loop.c (vect_is_slp_reduction): Delay reduction
2897 group building until we have successfully detected the SLP
2899 (vect_is_simple_reduction): Remove fixup code here.
2901 2018-11-13 David Malcolm <dmalcolm@redhat.com>
2903 * dumpfile.c (VERIFY_DUMP_ENABLED_P): New macro.
2904 (dump_gimple_stmt): Use it.
2905 (dump_gimple_stmt_loc): Likewise.
2906 (dump_gimple_expr): Likewise.
2907 (dump_gimple_expr_loc): Likewise.
2908 (dump_generic_expr): Likewise.
2909 (dump_generic_expr_loc): Likewise.
2910 (dump_printf): Likewise.
2911 (dump_printf_loc): Likewise.
2912 (dump_dec): Likewise.
2913 (dump_dec): Likewise.
2914 (dump_hex): Likewise.
2915 (dump_symtab_node): Likewise.
2916 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2917 Guard dump call with dump_enabled_p.
2918 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl): Likewise.
2919 * graphite-optimize-isl.c (optimize_isl): Likewise.
2920 * graphite.c (graphite_transform_loops): Likewise.
2921 * tree-loop-distribution.c (pass_loop_distribution::execute): Likewise.
2922 * tree-parloops.c (parallelize_loops): Likewise.
2923 * tree-ssa-loop-niter.c (number_of_iterations_exit): Likewise.
2924 * tree-vect-data-refs.c (vect_analyze_group_access_1): Likewise.
2925 (vect_prune_runtime_alias_test_list): Likewise.
2926 * tree-vect-loop.c (vect_update_vf_for_slp): Likewise.
2927 (vect_estimate_min_profitable_iters): Likewise.
2928 * tree-vect-slp.c (vect_record_max_nunits): Likewise.
2929 (vect_build_slp_tree_2): Likewise.
2930 (vect_supported_load_permutation_p): Likewise.
2931 (vect_slp_analyze_operations): Likewise.
2932 (vect_slp_analyze_bb_1): Likewise.
2933 (vect_slp_bb): Likewise.
2934 * tree-vect-stmts.c (vect_analyze_stmt): Likewise.
2935 * tree-vectorizer.c (try_vectorize_loop_1): Likewise.
2936 (pass_slp_vectorize::execute): Likewise.
2937 (increase_alignment): Likewise.
2939 2018-11-13 David Malcolm <dmalcolm@redhat.com>
2942 * ipa-inline.c (report_inline_failed_reason): Guard calls to
2943 cl_target_option_print_diff and cl_optimization_print_diff with
2946 2018-11-13 David Malcolm <dmalcolm@redhat.com>
2948 * doc/invoke.texi (-fsave-optimization-record): Note that the
2949 output is compressed.
2950 * optinfo-emit-json.cc: Include <zlib.h>.
2951 (optrecord_json_writer::write): Compress the output.
2953 2018-11-13 Aldy Hernandez <aldyh@redhat.com>
2955 * tree-vrp.c (value_range_base::dump): Dump type.
2956 Do not use INF nomenclature for 1-bit types.
2957 (dump_value_range): Group all variants to common dumping code.
2958 (debug): New overloaded functions for value_ranges.
2959 (value_range_base::dump): Remove no argument version.
2960 (value_range::dump): Same.
2962 2018-11-13 Richard Biener <rguenther@suse.de>
2964 PR tree-optimization/87931
2965 * tree-vect-loop.c (vect_is_simple_reduction): Restrict
2966 nested cycles we support to latch computations vectorizable_reduction
2969 2018-11-13 Martin Liska <mliska@suse.cz>
2971 PR tree-optimization/87885
2972 * cfghooks.c (account_profile_record): Rename
2974 (profile_record_check_consistency): ... this.
2975 Calculate missing num_mismatched_freq_in.
2976 (profile_record_account_profile): New function
2977 that calculates time and size of a function.
2978 * cfghooks.h (struct profile_record): Remove
2980 (struct cfg_hooks): Remove after_pass flag.
2981 (account_profile_record): Rename to ...
2982 (profile_record_check_consistency): ... this.
2983 (profile_record_account_profile): New.
2984 * cfgrtl.c (rtl_account_profile_record): Remove
2986 * passes.c (check_profile_consistency): Do only
2988 (account_profile): Calculate size and time of
2990 (pass_manager::dump_profile_report): Reformat
2992 (execute_one_ipa_transform_pass): Call
2993 consistency check before clean upand call account_profile
2994 after a clean up is done.
2995 (execute_one_pass): Call check_profile_consistency and
2996 account_profile instead of using after_pass flag..
2997 * tree-cfg.c (gimple_account_profile_record): Likewise.
2999 2018-11-13 Martin Liska <mliska@suse.cz>
3002 * config/i386/i386.c (ix86_option_override_internal): Error
3003 about usage -mabi=ms and -fsanitize={,kernel-}address.
3005 2018-11-13 Andre Vieira <andre.simoesdiasvieira@arm.com>
3007 * config/aarch64/aarch64.c
3008 (aarch64_vectorize_preferred_vector_alignment): Change return type to
3010 (aarch64_simd_vector_alignment_reachable): Adapt to preferred vector
3011 alignment being a poly int.
3012 * doc/tm.texi (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Change
3013 return type to poly_uint64.
3014 * target.def (default_preferred_vector_alignment): Likewise.
3015 * targhooks.c (default_preferred_vector_alignment): Likewise.
3016 * targhooks.h (default_preferred_vector_alignment): Likewise.
3017 * tree-vect-data-refs.c (vect_calculate_target_alignment): Likewise.
3018 (vect_compute_data_ref_alignment): Adapt to vector alignment being a
3020 (vect_update_misalignment_for_peel): Likewise.
3021 (vect_enhance_data_refs_alignment): Likewise.
3022 (vect_find_same_alignment_drs): Likewise.
3023 (vect_duplicate_ssa_name_ptr_info): Likewise.
3024 (vect_setup_realignment): Likewise.
3025 (vect_can_force_dr_alignment_p): Change alignment parameter type to
3027 * tree-vect-loop-manip.c (get_misalign_in_elems): Learn to construct a
3028 mask with a compile time variable vector alignment.
3029 (vect_gen_prolog_loop_niters): Adapt to vector alignment being a poly
3031 (vect_do_peeling): Exit early if vector alignment is not constant.
3032 * tree-vect-stmts.c (ensure_base_align): Adapt to vector alignment being
3034 (vectorizable_store): Likewise.
3035 (vectorizable_load): Likweise.
3036 * tree-vectorizer.h (struct dr_vec_info): Make target_alignment field a
3038 (vect_known_alignment_in_bytes): Adapt to vector alignment being a
3040 (vect_can_force_dr_alignment_p): Change alignment parameter type to
3043 2018-11-13 Richard Biener <rguenther@suse.de>
3045 PR tree-optimization/87962
3046 * tree-vect-loop.c (vect_is_simple_reduction): More reliably
3047 detect outer reduction for disqualifying in-loop uses.
3049 2018-11-13 Richard Biener <rguenther@suse.de>
3051 PR tree-optimization/87967
3052 * tree-vect-loop.c (vect_transform_loop): Also copy PHIs
3053 for constants for the scalar loop.
3055 2018-11-13 Alan Modra <amodra@gmail.com>
3057 * config/rs6000/linux64.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Exclude
3058 integer constants when -mcmodel=medium.
3060 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
3062 * config/arc/arc.h (FUNCTION_PROFILER): Redefine to empty.
3063 * config/arc/elf.h (PROFILE_HOOK): Define.
3064 * config/arc/linux.h (PROFILE_HOOK): Likewise.
3066 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
3068 * config/arc/arc.c (hwloop_optimize): Bailout when detecting a
3069 jump table data in the text section.
3071 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
3073 * config/arc/arc.c (arc_eh_return_address_location): Repurpose it
3074 to fit the eh_return pattern.
3075 * config/arc/arc.md (eh_return): Define.
3076 (VUNSPEC_ARC_EH_RETURN): Likewise.
3077 * config/arc/arc-protos.h (arc_eh_return_address_location): Match
3079 * config/arc/arc.h (EH_RETURN_HANDLER_RTX): Remove it.
3082 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
3084 * common/config/arc/arc-common.c (arc_option_optimization_table):
3085 Millicode optimization is default on for size optimizations.
3086 * config/arc/arc-protos.h (arc_check_multi): New function.
3087 * config/arc/arc.c (RTX_OK_FOR_OFFSET_P): Rearange.
3088 (ENTER_LEAVE_START_REG): Define.
3089 (ENTER_LEAVE_END_REG): Likewise.
3090 (arc_override_options): Disable millicode when long calls option
3092 (arc_frame_info): Change it from int to bool.
3093 (arc_compute_frame_size): Clean up.
3094 (arc_save_restore): Remove.
3095 (frame_save_reg): New function.
3096 (frame_restore_reg): Likewise.
3097 (arc_enter_leave_p): Likewise.
3098 (arc_save_callee_saves): Likewise.
3099 (arc_restore_callee_saves): Likewise.
3100 (arc_save_callee_enter): Likewise.
3101 (arc_restore_callee_leave): Likewise.
3102 (arc_save_callee_milli): Likewise.
3103 (arc_restore_callee_milli): Likewise.
3104 (arc_expand_prologue): Reimplement to emit enter/leave
3106 (arc_expand_epilogue): Likewise.
3107 (arc_check_multi): New function.
3108 * config/arc/arc.md (push_multi_fp): New pattern.
3109 (push_multi_fp_blink): Likewise.
3110 (pop_multi_fp): Likewise.
3111 (pop_multi_fp_blink): Likewise.
3112 (pop_multi_fp_ret): Likewise.
3113 (pop_multi_fp_blink_ret): Likewise.
3114 * config/arc/arc.opt (mmillicode): Update option.
3115 (mcode-density-frame): New option.
3116 * config/arc/predicates.md (push_multi_operand): New predicate.
3117 (pop_multi_operand): Likewise.
3118 * doc/invoke.texi (ARC): Update ARC options information.
3121 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
3123 * config/arc/arc-protos.h (gen_operands_ldd_std): Add.
3124 * config/arc/arc.c (operands_ok_ldd_std): New function.
3125 (mem_ok_for_ldd_std): Likewise.
3126 (gen_operands_ldd_std): Likewise.
3127 * config/arc/arc.md: Add peephole2 rules for std/ldd.
3129 2018-11-13 Eric Botcazou <ebotcazou@adacore.com>
3131 * toplev.c (output_stack_usage): Turn test on flag_stack_usage into
3132 test on stack_usage_file.
3133 (lang_dependent_init): Do not open the .su file if generating LTO.
3135 2018-11-13 Jakub Jelinek <jakub@redhat.com>
3137 PR rtl-optimization/87918
3138 * simplify-rtx.c (simplify_merge_mask): For COMPARISON_P, use
3139 simplify_gen_relational rather than simplify_gen_binary.
3141 2018-11-13 Richard Biener <rguenther@suse.de>
3143 * tree-ssanames.h (set_range_info): Use value_range_base.
3144 (get_range_info): Likewise.
3145 * tree-ssanames.c (set_range_info): Likewise.
3146 (get_range_info): Likewise.
3147 * tree-vrp.c (value_range_base::union_helper): Split
3148 out common parts of value_range[_base]::union_.
3149 (value_range_base::union_): Update.
3150 (value_range::union_): Likewise.
3151 (determine_value_range_1): Use value_range_base.
3152 (determine_value_range): Likewise.
3153 * tree-vrp.h (value_range_base::union_helper): Move ...
3154 (value_range::union_helper): ... from here.
3156 2018-11-13 Alan Modra <amodra@gmail.com>
3158 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Negate
3160 (rs6000_secondary_reload_gpr): Don't call find_replacement.
3162 2018-11-13 Sandra Loosemore <sandra@codesourcery.com>
3165 * doc/invoke.texi (Optimize Options): Clarify that the
3166 l1-cache-line-size, l1-cache-size, and l2-cache-size parameters
3167 apply to data cache size.
3169 2018-11-13 Alan Modra <amodra@gmail.com>
3171 * config/rs6000/predicates.md (logical_const_operand),
3172 (logical_operand): Correct comment.
3173 * config/rs6000/rs6000.c (rs6000_emit_prologue): Comment fix.
3175 2018-11-13 Alan Modra <amodra@gmail.com>
3177 * config/rs6000/rs6000.md (addsi3_high): Prefix with '*'.
3179 2018-11-13 Alan Modra <amodra@gmail.com>
3181 * gcc/doc/md.texi (Machine Constraints): Correct rs6000 h constraint
3183 * config/rs6000/rs6000.md (movsi_internal1): Delete MT%0 case
3184 covered by alternative.
3185 (movcc_internal1): Ignore h for register preference.
3186 (mov<mode>_hardfloat64): Likewise.
3187 (mov<mode>_softfloat): Ignore c, l, h for register preference.
3189 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
3191 PR preprocessor/47823
3192 * doc/cpp.texi (Alternatives to Wrapper #ifndef): Move #pragma once
3195 * doc/extend.texi (Pragmas): Note additional pragmas documented
3198 2018-11-13 Kugan Vivekanandarajah <kuganv@linaro.org>
3202 * tree-scalar-evolution.c (expression_expensive_p): Make
3203 BUILTIN POPCOUNT as expensive when backend does not define it.
3205 2018-11-12 Fredrik Noring <noring@nocrew.org>
3207 * config.gcc: Update with-llsc defaults for MIPS r5900.
3209 2018-11-12 Martin Liska <mliska@suse.cz>
3211 PR gcov-profile/87442
3212 * common.opt: Add -fprofile-filter-files and -fprofile-exclude-files
3214 * doc/invoke.texi: Document them.
3215 * tree-profile.c (parse_profile_filter): New.
3216 (parse_profile_file_filtering): Likewise.
3217 (release_profile_file_filtering): Likewise.
3218 (include_source_file_for_profile): Likewise.
3219 (tree_profiling): Filter source files based on the
3220 newly added options.
3222 2018-11-12 Martin Liska <mliska@suse.cz>
3225 * doc/extend.texi: Add missing values for __builtin_cpu_is and
3226 __builtin_cpu_supports for x86 target.
3228 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
3231 * doc/rtl.texi (Constants): Clarify that mode of "high" doesn't
3233 (Arithmetic): Likewise for "lo_sum".
3235 2018-11-12 Renlin Li <renlin.li@arm.com>
3238 * dse.c (get_stored_val): Add check for compile-time constantness
3241 2018-11-12 Sudakshina Das <sudi.das@arm.com>
3243 * config/arm/arm-cpus.in (armv8_5, sb, predres): New features.
3244 (ARMv8_5a): New fgroup.
3245 (armv8.5-a): New arch.
3246 (armv8-a, armv8.1-a, armv8.2-a, armv8.3-a, armv8.4-a): New options
3248 * config/arm/arm-tables.opt: Regenerate.
3249 * config/arm/t-aprofile: Add matching rules for -march=armv8.5-a
3250 * config/arm/t-arm-elf (all_v8_archs): Add armv8.5-a.
3251 * config/arm/t-multilib (v8_5_a_simd_variants): New variable.
3252 Add matching rules for -march=armv8.5-a and extensions.
3253 * doc/invoke.texi (ARM options): Document -march=armv8.5-a.
3254 Add sb and predres to all armv8-a except armv8.5-a.
3256 2018-11-12 Richard Biener <rguenther@suse.de>
3258 * tree-vrp.h (value_range[_base]::set): Make public. Provide
3259 overload for single value.
3260 (value_range[_base]::set_nonnull): New.
3261 (value_range[_base]::set_null): Likewise.
3262 (value_range): Document bitmap copying behavior, mark
3263 copy constructor and assignment operator deleted.
3264 (value_range::move): New.
3265 (value_range::set_and_canonicalize): Default bitmap to zero.
3266 (set_value_range_to_nonnull): Remove.
3267 (set_value_range_to_null): Likewise.
3268 (set_value_range): Likewise.
3269 (set_value_range_to_value): Likewise.
3270 (extract_range_from_unary_expr): Work on value_range_base.
3271 (extract_range_from_binary_expr_1): Likewise. Rename to...
3272 (extract_range_from_binary_expr): ... this.
3273 * tree-vrp.c (value_range::update): Clear equiv bitmap
3275 (value_range::move): New, move equiv bitmap.
3276 (value_range_base::set_undefined): Avoid assignment.
3277 (value_range::set_undefined): Likewise.
3278 (value_range_base::set_varying): Likewise.
3279 (value_range::set_varying): Likewise.
3280 (set_value_range): Remove.
3281 (value_range_base::set): New overload for value.
3282 (value_range::set): Likewise.
3283 (set_value_range_to_nonnull): Remove.
3284 (value_range_base::set_nonnull): New.
3285 (value_range::set_nonnull): Likewise.
3286 (set_value_range_to_null): Remove.
3287 (value_range_base::set_null): New.
3288 (value_range::set_null): Likewise.
3289 (range_is_null): Work on value_range_base.
3290 (range_is_nonnull): Likewise.
3291 (ranges_from_anti_range): Likewise.
3292 (extract_range_into_wide_ints): Likewise.
3293 (extract_range_from_multiplicative_op): Likewise.
3294 (extract_range_from_binary_expr): Likewise. Update for API changes.
3295 (extract_range_from_unary_expr): Likewise. Remove OBJ_TYPE_REF
3297 (value_range::intersect_helper): Avoid copy and assignment.
3298 (value_range::union_helper): Likewise.
3299 (determine_value_range_1): Adjust.
3300 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
3301 Avoid assignment by using move.
3302 (evrp_range_analyzer::record_ranges_from_stmt): Avoid assignment.
3303 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
3305 * tree-ssanames.c (get_range_info): Likewise.
3306 * vr-values.h (vr_values::get_vr_for_comparison): Adjust API.
3307 * vr-values.c (vr_values::get_value_range): Adjust.
3308 (vr_values::update_value_range): Likewise.
3309 (symbolic_range_based_on_p): Work on value_range_base.
3310 (vr_values::extract_range_from_binary_expr): Use value_range_base.
3311 (vr_values::extract_range_from_unary_expr): Likewise.
3312 (vr_values::extract_range_from_cond_expr): Avoid assignment.
3313 (vr_values::extract_range_from_comparison): Adjust.
3314 (vr_values::check_for_binary_op_overflow): Use value_range_base.
3315 (vr_values::extract_range_basic): Adjust.
3316 (vr_values::adjust_range_with_scev): Likewise.
3317 (vr_values::vrp_visit_assignment_or_call): Likewise.
3318 (vr_values::get_vr_for_comparison): Change API to avoid
3319 assignment and copy construction.
3320 (vr_values::compare_name_with_value): Adjust accordingly.
3321 (vr_values::compare_names): Likewise.
3322 (vr_values::extract_range_from_phi_node): Avoid assignment and
3323 bogus in-place modify of equiv bitmap.
3324 (vr_values::simplify_bit_ops_using_ranges): Use value_range_base.
3325 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Adjust
3326 for extract_range_from_unary_expr API change.
3327 * ipa-cp.c (ipa_vr_operation_and_type_effects): Likewise.
3329 2018-11-12 Eric Botcazou <ebotcazou@adacore.com>
3331 * config/mcore/mcore.h (WORD_REGISTER_OPERATIONS): Remove duplicate.
3333 2018-11-12 Richard Biener <rguenther@suse.de>
3335 * tree-vrp.h (value_range_base::symbolic_p,
3336 value_range_base::constant_p, value_range_base::zero_p,
3337 value_range_base::singleton_p): Move from value_range.
3338 (value_range::dump): Add.
3339 * gimple-ssa-evrp-analyze.c
3340 (evrp_range_analyzer::record_ranges_from_phis): Use set_varying.
3341 * ipa-cp.c (ipcp_vr_lattice::print): Use dump_value_range.
3342 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
3344 * tree-vrp.c (value_range::symbolic_p): Move to value_range_base.
3345 (value_range::constant_p): Likewise.
3346 (value_range::singleton_p): Likewise.
3347 (value_range_base::dump): Add.
3348 (set_value_range_to_undefined): Remove.
3349 (set_value_range_to_varying): Likewise.
3350 (range_int_cst_p): Take value_range_base argument.
3351 (range_int_cst_singleton_p): Likewise.
3352 (value_range_constant_singleton): Likewise.
3353 (vrp_set_zero_nonzero_bits): Likewise.
3354 (extract_range_from_multiplicative_op): Use set_varying.
3355 (extract_range_from_binary_expr_1): Likewise. Use set_undefined.
3356 (extract_range_from_unary_expr): Likewise.
3357 (dump_value_range_base): Change to overload of dump_value_range.
3358 (vrp_prop::vrp_initialize): Use set_varying and set_undefined.
3359 (vrp_prop::visit_stmt): Likewise.
3360 (value_range::intersect_helper): Likewise.
3361 (value_range::union_helper): Likewise.
3362 (determine_value_range_1): Likewise.
3364 2018-11-12 Richard Biener <rguenther@suse.de>
3366 * tree-vrp.c (set_value_range_to_nonnull): Clear equiv.
3367 (set_value_range_to_null): Likewise.
3368 * vr-values.c (vr_values::extract_range_from_comparison):
3369 Clear equiv for constant singleton ranges.
3371 2018-11-12 Wei Xiao <wei3.xiao@intel.com>
3373 * config/i386/sse.md: Combine VFIXUPIMM* patterns
3374 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
3375 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
3376 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Remove.
3377 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
3378 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
3379 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Remove.
3381 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
3384 * doc/extend.texi (Common Type Attributes): For the align type
3385 attribute, copy language about decreasing alignment from the
3386 corresponding variable attribute.
3388 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
3390 * config/csky/csky.md (*fpuv2_nmulsf3_1, *fpuv2_nmuldf3_1): Handle
3393 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
3396 * doc/invoke.texi (C++ Dialect Options): Add warning about mixing
3397 -frtti and -fno-rtti code.
3399 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
3402 * doc/extend.texi (Other Builtins): Document probability associated
3403 with __builtin_expect.
3405 2018-11-11 Uros Bizjak <ubizjak@gmail.com>
3408 * config/i386/i386.h (STACK_BOUNDARY): Use TARGET_64BIT_MS_ABI
3409 instead of (TARGET_64BIT && ix86_abi == MS_ABI).
3410 * config/i386/darwin.h (STACK_BOUNDARY): Ditto.
3411 * config/i386/cygming.h (STACK_BOUNDARY): Remove.
3413 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
3415 * config/csky/csky-linux-elf.h (CC1_SPEC): Support -profile.
3417 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
3419 * config/csky/csky.h (ASM_PREFERRED_EH_DATA_FORMAT): Define.
3421 2018-11-11 Richard Biener <rguenther@suse.de>
3423 * tree-vrp.h (class value_range_base): New base class for
3424 value_range containing all but the m_equiv member.
3425 (dump_value_range_base): Add.
3426 (range_includes_zero_p): Work on value_range_base.
3427 * tree-vrp.c (value_range_base::set): Split out base handling
3429 (value_range::set): this.
3430 (value_range::set_equiv): New.
3431 (value_range_base::value_range_base): New constructors.
3432 (value_range_base::check): Split out base handling from...
3433 (value_range::check): this.
3434 (value_range::equal_p): Refactor in terms of
3435 ignore_equivs_equal_p which is now member of the base.
3436 (value_range_base::set_undefined): New.
3437 (value_range_base::set_varying): Likewise.
3438 (value_range_base::dump):Split out base handling from...
3439 (value_range::dump): this.
3440 (value_range_base::set_and_canonicalize): Split out base handling
3442 (value_range::set_and_canonicalize): this.
3443 (value_range_base::union_): New.
3444 * ipa-prop.h (struct ipa_jump_func): Use value_range_base *
3446 * ipa-cp.c (class ipcp_vr_lattice): Use value_range_base
3447 instead of value_range everywhere.
3448 (ipcp_vr_lattice::print): Use dump_value_range_base.
3449 (ipcp_vr_lattice::meet_with): Adjust.
3450 (ipcp_vr_lattice::meet_with_1): Likewise.
3451 (ipa_vr_operation_and_type_effects): Likewise.
3452 (propagate_vr_across_jump_function): Likewise.
3453 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Likewise.
3454 (ipa_get_value_range): Likewise.
3455 (ipa_set_jfunc_vr): Likewise.
3456 (ipa_compute_jump_functions_for_edge): Likewise.
3458 2018-11-10 Sandra Loosemore <sandra@codesourcery.com>
3461 * doc/invoke.texi (Optimize Options): Add @opindex entries
3462 for the positive forms of -fno-xxx and -mno-xxx options
3463 that were lacking them.
3465 2018-11-10 Segher Boessenkool <segher@kernel.crashing.org>
3467 * combine.c (make_more_copies): Only make an intermediate copy if the
3468 dest of a move is a pseudo.
3470 2018-11-09 Maya Rashish <coypu@sdf.org>
3473 * config/netbsd-elf.h (NETBSD_STARTFILE_SPEC): Use crtbeginS.o for PIE.
3474 (NETBSD_ENDFILE_SPEC): Use crtendS.o for PIE.
3476 2018-11-09 Sandra Loosemore <sandra@codesourcery.com>
3480 * doc/invoke.texi (Optimize Options): Clarify default behavior
3481 for -fno-toplevel-reorder, -fno-defer-pop, and -fno-branch-count-reg.
3483 2018-11-09 Bernd Edlinger <bernd.edlinger@hotmail.de>
3485 PR tree-optimization/87940
3486 * expr.c (string_constant): Don't strip NOPS in subexpressions.
3487 Fold PLUS_EXPR correctly.
3489 2018-11-09 Ilya Leoshkevich <iii@linux.ibm.com>
3492 * config/s390/s390.c (s390_safe_relative_long_p): New function.
3493 (annotate_constant_pool_refs): Skip insns which support
3494 relative addressing.
3495 (annotate_constant_pool_refs_1): New helper function.
3496 (find_constant_pool_ref): Skip insns which support relative
3498 (find_constant_pool_ref_1): New helper function.
3499 (replace_constant_pool_ref): Skip insns which support
3500 relative addressing.
3501 (replace_constant_pool_ref_1): New helper function.
3502 (s390_mainpool_start): Adapt to the new signature.
3503 (s390_mainpool_finish): Likewise.
3504 (s390_chunkify_start): Likewise.
3505 (s390_chunkify_finish): Likewise.
3506 (pass_s390_early_mach::execute): Likewise.
3507 (s390_prologue_plus_offset): Likewise.
3508 (s390_emit_prologue): Likewise.
3509 (s390_emit_epilogue): Likewise.
3511 2018-11-09 Jakub Jelinek <jakub@redhat.com>
3513 * gimplify.c (gimplify_scan_omp_clauses): Call sorry_at for valid
3514 but unsupported lastprivate with conditional modifier.
3516 2018-11-09 Jeff Law <law@redhat.com>
3518 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): Handle
3519 unused argument better. Add gcc_unreachable to silence warning.
3521 2018-11-09 Martin Sebor <msebor@redhat.com>
3524 * attribs.c (has_attribute): New helper function.
3525 (decls_mismatched_attributes, maybe_diag_alias_attributes): Same.
3526 * attribs.h (decls_mismatched_attributes): Declare.
3527 * cgraphunit.c (handle_alias_pairs): Call maybe_diag_alias_attributes.
3528 (maybe_diag_incompatible_alias): Use OPT_Wattribute_alias_.
3529 * common.opt (-Wattribute-alias): Take an argument.
3530 (-Wno-attribute-alias): New option.
3531 * doc/extend.texi (Common Function Attributes): Document copy.
3532 (Common Variable Attributes): Same.
3533 * doc/invoke.texi (-Wmissing-attributes): Document enhancement.
3534 (-Wattribute-alias): Document new option argument.
3536 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
3538 * config/arm/parsecpu.awk (/alias/): Tighten invisible alias
3539 matching criteria. Remove unused array initializer.
3541 2018-11-09 Bill Schmidt <wschmidt@linux.ibm.com>
3542 Jinsong Ji <jji@us.ibm.com>
3544 * config/rs6000/xmmintrin.h (_mm_cvtss_si32): Fix incorrect
3545 constraints by introducing a new temporary.
3546 (_mm_cvtss_si64): Likewise.
3548 2018-11-09 Martin Liska <mliska@suse.cz>
3550 * common.opt: Add -fipa-stack-alignment flag.
3551 * doc/invoke.texi: Document it.
3552 * final.c (rest_of_clean_state): Guard stack
3553 shrinking with flag.
3555 2018-11-09 Martin Liska <mliska@suse.cz>
3557 * cgraph.h (ipa_discover_readonly_nonaddressable_vars): Rename
3559 (ipa_discover_variable_flags): ... this.
3560 * common.opt: Come up with new flag -fipa-reference-addressable.
3561 * doc/invoke.texi: Document it.
3562 * ipa-reference.c (propagate): Call the renamed fn.
3563 * ipa-visibility.c (whole_program_function_and_variable_visibility):
3565 * ipa.c (ipa_discover_readonly_nonaddressable_vars): Renamed to
3567 (ipa_discover_variable_flags): ... this. Discover
3568 non-addressable variables only with the newly added flag.
3569 * opts.c: Enable the newly added flag with -O1 and higher
3572 2018-11-09 David Malcolm <dmalcolm@redhat.com>
3574 * json.cc (selftest::test_writing_literals): Fix comment.
3576 2018-11-09 Martin Liska <mliska@suse.cz>
3578 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not copy
3579 string to a stack buffer.
3580 (aarch64_parse_cpu): Likewise.
3581 (aarch64_parse_tune): Likewise.
3583 2018-11-09 Richard Biener <rguenther@suse.de>
3585 PR tree-optimization/87953
3586 * tree-vect-loop.c (vectorizable_reduction): For analysis
3587 always pass ops[0] to vectorizable_condition.
3589 2018-11-09 Stafford Horne <shorne@gmail.com>
3590 Richard Henderson <rth@twiddle.net>
3591 Joel Sherrill <joel@rtems.org>
3593 * common/config/or1k/or1k-common.c: New file.
3594 * config/or1k/*: New.
3595 * config.gcc (or1k*-*-*): New.
3596 * configure.ac (or1k*-*-*): New test for openrisc tls.
3597 * configure: Regenerated.
3598 * doc/install.texi: Document OpenRISC triplets.
3599 * doc/invoke.texi: Document OpenRISC arguments.
3600 * doc/md.texi: Document OpenRISC.
3602 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
3604 * config/arm/arm-cpus.in (arm7tdmi): Add an alias for arm7tdmi-s.
3605 (arm7tdmi-s): Delete CPU.
3606 (arm710t): Add aliases for arm720t and arm740t.
3607 (arm720t, arm740t): Delete CPUs.
3608 (arm920t): Add aliases for arm920, arm922t and arm940t.
3609 (arm920, arm922t, arm940t): Delete CPUs.
3610 (arm10tdmi): Add alias for arm1020t.
3611 (arm1020t): Delete CPU.
3612 (arm9e): Add aliases for arm946e-s, arm966e-s and arm968e-s.
3613 (arm946e-s, arm966e-s, arm968e-s): Delete CPUs.
3614 (arm10e): Add aliases for arm1020e and arm1022e.
3615 (arm1020e, arm1022e): Delete CPU.
3616 * config/arm/arm.md (generic_sched): Remove entries that are now
3618 (generic_vfp): Likewise.
3619 * config/arm/arm1020e.md: Simplify tuning selection based on alias
3621 * config/arm/arm-tune.md: Regenerated.
3622 * config/arm/arm-tables.opt: Regenerated.
3624 2018-11-09 Richard Biener <rguenther@suse.de>
3626 PR tree-optimization/87621
3627 * tree-vect-loop.c (vectorizable_reduction): Handle reduction
3628 op with only phi inputs.
3629 * tree-ssa-loop-ch.c: Include tree-ssa-sccvn.h.
3630 (ch_base::copy_headers): Run CSE on copied loop headers.
3631 (pass_ch_vect::process_loop_p): Simplify.
3633 2018-11-09 Alexandre Oliva <oliva@adacore.com>
3635 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Adjust
3636 for non-w64 x86_64 biarch.
3638 2018-11-09 Alexandre Oliva <aoliva@redhat.com>
3640 PR rtl-optimization/86438
3641 * compare-elim.c (try_eliminate_compare): Use SET_SRC instead
3642 of in_b for the compare if in_b is SET_DEST.
3645 * config/i386/i386.c (ix86_const_not_ok_for_debug_p): Reject
3646 non-toplevel UNSPEC.
3648 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
3650 * tree-vrp.c (value_range::check): Do not access internals
3652 (value_range::singleton_p): Same.
3653 (value_range::type): Same.
3654 (vrp_finalize): Use value_range API.
3656 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
3658 * tree-vrp.c (may_contain_p): Do not access m_min/m_max directly.
3660 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
3662 * vr-values.c (vr_values::get_value_range): Use value_range API
3663 instead of piecing together ranges.
3664 (vr_values::update_value_range): Same.
3666 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
3668 * gimple-fold.c (size_must_be_zero_p): Use value_range API instead
3669 of performing ad-hoc calculations.
3670 * tree-ssanames.c (set_range_info): New overloaded function
3671 accepting value_range &.
3672 (get_range_info): Same.
3673 * tree-ssanames.h (set_range_info_raw): Remove.
3674 (set_range_info): New prototype.
3675 (get_range_info): Same.
3676 * tree-vrp.h (value_range::null_p): Rename to zero_p.
3677 * tree-vrp.c (value_range::null_p): Same.
3679 2018-11-09 Jan Hubicka <jh@suse.cz>
3681 * tree.c (fld_type_variant_equal_p): Test user align flag.
3682 (flt_type_variant): Copy user align flag.
3683 (fld_incomplete_type_of): Clear it.
3685 2018-11-09 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3687 * config/arm/neon.md (div<mode>3): New pattern.
3689 2018-11-08 Andi Kleen <ak@linux.intel.com>
3691 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PTWRITE_SET): New.
3692 (OPTION_MASK_ISA_PTWRITE_UNSET): New.
3693 (ix86_handle_option): Handle OPT_mptwrite.
3694 * config/i386/cpuid.h (bit_PTWRITE): Add.
3695 * config/i386/driver-i386.c (host_detect_local_cpu): Detect ptwrite.
3696 * config/i386/i386-builtin.def (BDESC): Add ptwrite32/64.
3697 * config/i386/i386-c.c (ix86_target_macros_internal): Define __PTWRITE__.
3698 * config/i386/i386.c (ix86_target_string): Handle ptwrite.
3699 (ix86_option_override_internal): Handle PTA_PTWRITE.
3700 (ix86_valid_target_attribute_inner_p): Define ptwrite.
3701 (def_builtin2): Force UINT64 to be 64bit only.
3702 * config/i386/i386.h (TARGET_PTWRITE): Add.
3703 (TARGET_PTWRITE_P): Add.
3705 * config/i386/i386.md: Define ptwrite.
3706 * config/i386/i386.opt: Add -mptwrite.
3707 * config/i386/immintrin.h (_ptwrite64): Add.
3709 * doc/extend.texi: Document __builtin_ia32_ptwrite*.
3710 * doc/invoke.texi: Document -mptwrite.
3712 2018-11-08 Peter Bergner <bergner@linux.ibm.com>
3714 PR rtl-optimization/87600
3715 * cfgexpand.c (expand_asm_stmt): Catch illegal asm constraint usage.
3716 * lra-constraints.c (process_alt_operands): Skip illegal hard
3717 register usage. Prefer reloading non hard register operands.
3719 2018-11-08 Sandra Loosemore <sandra@codesourcery.com>
3722 * doc/invoke.texi (Optimize Options): Clarify default behavior
3723 for -fno-sched-interblock and -fno-sched-spec.
3725 2018-11-08 Roman Geissler <roman.geissler@amadeus.com>
3727 * collect2.c (linker_select): Add USE_LLD_LD.
3728 (ld_suffixes): Add ld.lld.
3729 (main): Handle -fuse-ld=lld.
3730 * common.opt (-fuse-ld=lld): New option.
3731 * doc/invoke.texi (-fuse-ld=lld): Document.
3732 * opts.c (common_handle_option): Handle OPT_fuse_ld_lld.
3734 2018-11-08 Paul Koning <ni1d@arrl.net>
3736 * config/pdp11/constraints.md: Add "Z" series constraints for use
3737 with pre-dec and post-inc addressing.
3738 * config/pdp11/pdp11-protos.m (expand_block_move): Delete.
3739 (pdp11_expand_operands): Add int argument (word count).
3740 (pdp11_sp_frame_offset): Delete.
3741 (pdp11_cmp_length): New function.
3742 (pushpop_regeq): New function.
3743 * config/pdp11/pdp11.c (TARGET_STACK_PROTECT_RUNTIME_ENABLED_P):
3745 (pdp11_expand_prologue, pdp11_expand_epilogue): Rewrite for new
3747 (pdp11_initial_elimination_offset): Ditto.
3748 (pdp11_expand_operands): Add word count argument. Bugfixes.
3749 (output_move_multiple): Change how pointer adjustment is done.
3750 (pdp11_gen_int_label): Correct format.
3751 (output_ascii): Ditto.
3752 (pdp11_asm_output_var): Add code for DEC assembler case.
3753 (pdp11_asm_print_operand): Bugfix for CONST_DOUBLE holding integer
3755 (legitimate_const_double_p): Ditto.
3756 (pdp11_register_move_cost): Adjust for new register classes.
3757 (pdp11_regno_reg_class): Ditto.
3758 (expand_block_move): Delete.
3759 (pushpop_regeq): New function.
3760 (pdp11_legitimate_address_p): Bugfix in check for constant
3762 (pdp11_sp_frame_offset): Delete.
3763 (pdp11_reg_save_size): New helper function for new frame layout.
3764 (output_addr_const_pdp11): Remove CONST_DOUBLE case.
3765 (pdp11_expand_shift): Bugfix in check for constant shift count.
3766 (pdp11_shift_length): Ditto.
3767 (pdp11_assemble_shift): Copy input to pdp11_expand_operands.
3768 (pdp11_cmp_length): New function.
3769 * config/pdp11/pdp11.h (TARGET_CPU_CPP_BUILTINS): Add macros for
3770 some compile options.
3771 (FIXED_REGISTERS): Remove HARD_FRAME_POINTER_REGNUM.
3772 (CALL_USED_REGISTERS): Ditto.
3773 (ELIMINABLE_REGS): Ditto.
3774 (REGISTER_NAMES): Ditto.
3775 (reg_class): Add classes NOTR0_REG through NOTSP_REG for use by Z
3777 (REG_CLASS_NAMES): Ditto.
3778 (REG_CLASS_CONTENTS): Ditto. Also remove
3779 HARD_FRAME_POINTER_REGNUM.
3780 (CPU_REG_CLASS): New macro.
3781 (CLASS_MAX_NREGS): Adjust for new register classes.
3782 (FUNCTION_PROFILER): Make no-op.
3783 (may_call_alloca): Remove unused declaration.
3784 (ASM_OUTPUT_ALIGN): Add workaround for PR87795.
3785 (ASM_OUTPUT_SKIP): Fix format.
3786 * config/pdp11/pdp11.md (unspecv): Add UNSPECV_MOVMEM.
3787 (HARD_FRAME_POINTER_REGNUM): Remove.
3789 (*rts): Rename. Remove epilogue related checks.
3790 (cmpsi, cmpdi): New insn.
3791 (cbranch<mode>4): Change to apply to SI and DI modes as well.
3792 (mov<mode>): Change constraints to enforce that push/pop
3793 destination cannot use the same register as source.
3794 (*mov<mode><cc_cc>): Ditto.
3795 (movmemhi, movmemhi1, movmemhi_nocc): Change to expand block move
3796 at assembly output rather than as RTL expander.
3797 (zero_extendqihi2): Bugfix in check for same registers.
3798 (adddi3_nocc): Bugfix in check for constant operand.
3799 (addsi3_nocc): Ditto.
3800 (subdi3_nocc): Ditto.
3801 (subsi3_nocc): Ditto.
3802 (negdi2_nocc): Copy input to pdp11_expand_operands.
3803 (negsi2_nocc): Ditto.
3804 (bswap2_nocc): Ditto.
3805 * config/pdp11/pdp11.opt (mlra): Fix documentation.
3806 * config/pdp11/t-pdp11: Use -Os.
3808 2018-11-08 Richard Earnshaw <rearnsha@arm.com>
3810 * config/arm/parsecpu.awk (/alias/): New parsing rule.
3811 (/begin cpu/): Check that the cpu name hasn't been previously defined.
3812 (gen_comm_data): Print out CPU alias tables.
3813 (check_cpu): Match aliases when checking the CPU name.
3814 * config/arm/arm-protos.h (cpu_alias): New structure.
3815 (cpu_option): Add entry for aliases.
3816 * config/arm/arm-cpus.in (strongarm): Add aliases for strongarm110
3817 strongarm1100 and strongarm1110.
3818 (strongarm110, strongarm1100, strongarm1110): Delete CPU entries.
3819 (config/arm/arm-generic.md): Remove redundant references to
3820 strongarm110, strongarm1100 and strongarm1110.
3821 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
3822 Scan aliases for additional hints.
3823 (arm_parse_cpu_option_name): Also match a cpu name against the list
3825 * config/arm/arm-tables.opt: Regenerated.
3826 * config/arm/arm-tune.md: Regenerated.
3828 2018-11-08 Jakub Jelinek <jakub@redhat.com>
3830 * builtin-types.def (BT_FN_VOID_BOOL, BT_FN_VOID_SIZE_SIZE_PTR,
3831 BT_FN_UINT_UINT_PTR_PTR, BT_FN_UINT_OMPFN_PTR_UINT_UINT,
3832 BT_FN_BOOL_UINT_LONGPTR_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
3833 BT_FN_BOOL_UINT_ULLPTR_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR,
3834 BT_FN_BOOL_LONG_LONG_LONG_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
3835 BT_FN_BOOL_BOOL_ULL_ULL_ULL_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR): New.
3836 * gengtype.c (open_base_files): Add omp-general.h.
3837 * gimple.c (gimple_build_omp_critical):
3838 (gimple_build_omp_taskgroup): Add CLAUSES argument. Call
3839 gimple_omp_taskgroup_set_clauses.
3840 (gimple_build_omp_atomic_load): Add mo argument, call
3841 gimple_omp_atomic_set_memory_order.
3842 (gimple_build_omp_atomic_store): Likewise.
3843 (gimple_copy): Adjust handling of GIMPLE_OMP_TASKGROUP.
3844 * gimple.def (GIMPLE_OMP_TASKGROUP): Use GSS_OMP_SINGLE_LAYOUT
3846 (GIMPLE_OMP_TEAMS): Use GSS_OMP_PARALLEL_LAYOUT instead
3847 of GSS_OMP_SINGLE_LAYOUT, adjust comments.
3848 * gimple.h (enum gf_mask): Add GF_OMP_TEAMS_HOST, GF_OMP_TASK_TASKWAIT
3849 and GF_OMP_ATOMIC_MEMORY_ORDER. Remove GF_OMP_ATOMIC_SEQ_CST, use
3850 different value for GF_OMP_ATOMIC_NEED_VALUE.
3851 (struct gimple_statement_omp_taskreg): Add GIMPLE_OMP_TEAMS to
3853 (struct gimple_statement_omp_single_layout): And remove here.
3854 (struct gomp_teams): Inherit from gimple_statement_omp_taskreg rather
3855 than gimple_statement_omp_single_layout.
3856 (is_a_helper <gimple_statement_omp_taskreg *>::test): Allow
3858 (is_a_helper <const gimple_statement_omp_taskreg *>::test): Likewise.
3859 (gimple_omp_subcode): Formatting fix.
3860 (gimple_omp_teams_child_fn, gimple_omp_teams_child_fn_ptr,
3861 gimple_omp_teams_set_child_fn, gimple_omp_teams_data_arg,
3862 gimple_omp_teams_data_arg_ptr, gimple_omp_teams_set_data_arg,
3863 gimple_omp_teams_host, gimple_omp_teams_set_host,
3864 gimple_omp_task_taskwait_p, gimple_omp_task_set_taskwait_p,
3865 gimple_omp_taskgroup_clauses, gimple_omp_taskgroup_clauses_ptr,
3866 gimple_omp_taskgroup_set_clauses): New inline functions.
3867 (gimple_build_omp_atomic_load): Add enum omp_memory_order argument.
3868 (gimple_build_omp_atomic_store): Likewise.
3869 (gimple_omp_atomic_seq_cst_p): Remove.
3870 (gimple_omp_atomic_memory_order): New function.
3871 (gimple_omp_atomic_set_seq_cst): Remove.
3872 (gimple_omp_atomic_set_memory_order): New function.
3873 (gimple_build_omp_taskgroup): Add clauses argument.
3874 * gimple-pretty-print.c (dump_gimple_omp_taskgroup): New function.
3875 (dump_gimple_omp_task): Print taskwait with depend clauses.
3876 (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Use
3877 dump_omp_atomic_memory_order.
3878 (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP.
3879 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP_ALLOC_ONLY,
3880 GOVD_MAP_FROM_ONLY and GOVD_NONTEMPORAL.
3881 (enum omp_region_type): Reserve bits 1 and 2 for auxiliary flags,
3882 renumber values of most of ORT_* enumerators, add ORT_HOST_TEAMS,
3883 ORT_COMBINED_HOST_TEAMS, ORT_TASKGROUP, ORT_TASKLOOP and
3884 ORT_UNTIED_TASKLOOP enumerators.
3885 (enum gimplify_defaultmap_kind): New.
3886 (struct gimplify_omp_ctx): Remove target_map_scalars_firstprivate and
3887 target_map_pointers_as_0len_arrays members, add defaultmap.
3888 (new_omp_context): Initialize defaultmap member.
3889 (gimple_add_tmp_var): Handle ORT_TASKGROUP like ORT_WORKSHARE.
3890 (maybe_fold_stmt): Don't fold even in host teams regions.
3891 (omp_firstprivatize_variable): Handle ORT_TASKGROUP like
3892 ORT_WORKSHARE. Test ctx->defaultmap[GDMK_SCALAR] instead of
3893 ctx->omp_firstprivatize_variable.
3894 (omp_add_variable): Don't add private/firstprivate for VLAs in
3896 (omp_default_clause): Print "taskloop" rather than "task" if
3898 (omp_notice_variable): Handle ORT_TASKGROUP like ORT_WORKSHARE.
3899 Handle new defaultmap clause kinds.
3900 (omp_is_private): Handle ORT_TASKGROUP like ORT_WORKSHARE. Allow simd
3901 iterator to be lastprivate or private. Fix up diagnostics if linear
3902 is used on collapse>1 simd iterator.
3903 (omp_check_private): Handle ORT_TASKGROUP like ORT_WORKSHARE.
3904 (gimplify_omp_depend): New function.
3905 (gimplify_scan_omp_clauses): Add shared clause on parallel for
3906 combined parallel master taskloop{, simd} if taskloop has
3907 firstprivate, lastprivate or reduction clause. Handle
3908 OMP_CLAUSE_REDUCTION_TASK diagnostics. Adjust tests for
3909 ORT_COMBINED_TEAMS. Gimplify depend clauses with iterators. Handle
3910 cancel and simd OMP_CLAUSE_IF_MODIFIERs. Handle
3911 OMP_CLAUSE_NONTEMPORAL. Handle new defaultmap clause kinds. Handle
3912 OMP_CLAUSE_{TASK,IN}_REDUCTION. Diagnose invalid conditional
3914 (gimplify_adjust_omp_clauses_1): Ignore GOVD_NONTEMPORAL. Handle
3915 GOVD_MAP_ALLOC_ONLY and GOVD_MAP_FROM_ONLY.
3916 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_NONTEMPORAL. Handle
3917 OMP_CLAUSE_{TASK,IN}_REDUCTION.
3918 (gimplify_omp_task): Handle taskwait with depend clauses.
3919 (gimplify_omp_for): Add shared clause on parallel for combined
3920 parallel master taskloop{, simd} if taskloop has firstprivate,
3921 lastprivate or reduction clause. Use ORT_TASKLOOP or
3922 ORT_UNTIED_TASKLOOP instead of ORT_TASK or ORT_UNTIED_TASK. Adjust
3923 tests for ORT_COMBINED_TEAMS. Handle C++ range for loops with
3924 NULL TREE_PURPOSE in OMP_FOR_ORIG_DECLS. Firstprivatize
3925 __for_end and __for_range temporaries on OMP_PARALLEL for
3926 distribute parallel for{, simd}. Move OMP_CLAUSE_REDUCTION
3927 and OMP_CLAUSE_IN_REDUCTION from taskloop to the task construct
3928 sandwiched in between two taskloops.
3929 (computable_teams_clause): Test ctx->defaultmap[GDMK_SCALAR]
3930 instead of ctx->omp_firstprivatize_variable.
3931 (gimplify_omp_workshare): Set ort to ORT_HOST_TEAMS or
3932 ORT_COMBINED_HOST_TEAMS if not inside of target construct. If
3933 host teams, use gimplify_and_return_first etc. for body like
3934 for target or target data constructs, and at the end call
3935 gimple_omp_teams_set_host on the GIMPLE_OMP_TEAMS object.
3936 (gimplify_omp_atomic): Use OMP_ATOMIC_MEMORY_ORDER instead
3937 of OMP_ATOMIC_SEQ_CST, pass it as new argument to
3938 gimple_build_omp_atomic_load and gimple_build_omp_atomic_store, remove
3939 gimple_omp_atomic_set_seq_cst calls.
3940 (gimplify_expr) <case OMP_TASKGROUP>: Move handling into a separate
3941 case, handle taskgroup clauses.
3942 * lto-streamer-out.c (hash_tree): Handle
3943 OMP_CLAUSE_{TASK,IN}_REDUCTION.
3944 * Makefile.in (GTFILES): Add omp-general.h.
3945 * omp-builtins.def (BUILT_IN_GOMP_TASKWAIT_DEPEND,
3946 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_START,
3947 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_START,
3948 BUILT_IN_GOMP_LOOP_START, BUILT_IN_GOMP_LOOP_ORDERED_START,
3949 BUILT_IN_GOMP_LOOP_DOACROSS_START,
3950 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_NEXT,
3951 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
3952 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_START,
3953 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_START,
3954 BUILT_IN_GOMP_LOOP_ULL_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_START,
3955 BUILT_IN_GOMP_LOOP_ULL_DOACROSS_START,
3956 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_NEXT,
3957 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
3958 BUILT_IN_GOMP_PARALLEL_LOOP_NONMONOTONIC_RUNTIME,
3959 BUILT_IN_GOMP_PARALLEL_LOOP_MAYBE_NONMONOTONIC_RUNTIME,
3960 BUILT_IN_GOMP_PARALLEL_REDUCTIONS, BUILT_IN_GOMP_SECTIONS2_START,
3961 BUILT_IN_GOMP_TEAMS_REG, BUILT_IN_GOMP_TASKGROUP_REDUCTION_REGISTER,
3962 BUILT_IN_GOMP_TASKGROUP_REDUCTION_UNREGISTER,
3963 BUILT_IN_GOMP_TASK_REDUCTION_REMAP,
3964 BUILT_IN_GOMP_WORKSHARE_TASK_REDUCTION_UNREGISTER): New builtins.
3965 * omp-expand.c (workshare_safe_to_combine_p): Return false for
3966 non-worksharing loops.
3967 (omp_adjust_chunk_size): Don't adjust anything if chunk_size is zero.
3968 (determine_parallel_type): Don't combine parallel with worksharing
3969 which has _reductemp_ clause.
3970 (expand_parallel_call): Emit the GOMP_*nonmonotonic_runtime* or
3971 GOMP_*maybe_nonmonotonic_runtime* builtins instead of GOMP_*runtime*
3972 if there is nonmonotonic modifier or if there is no modifier and no
3973 ordered clause. For dynamic and guided schedule without monotonic
3974 and nonmonotonic modifier, default to nonmonotonic.
3975 (expand_omp_for): Likewise. Adjust expand_omp_for_generic caller, use
3976 GOMP_loop{,_ull}{,_ordered,_doacross}_start builtins if there are
3978 (expand_task_call): Add GOMP_TASK_FLAG_REDUCTION flag to flags if
3979 there are any reduction clauses.
3980 (expand_taskwait_call): New function.
3981 (expand_teams_call): New function.
3982 (expand_omp_taskreg): Allow GIMPLE_OMP_TEAMS and call
3983 expand_teams_call for it. Formatting fix. Handle taskwait with
3985 (expand_omp_for_generic): Add SCHED_ARG argument. Handle expansion
3986 of worksharing loops with task reductions.
3987 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk): Handle
3988 expansion of worksharing loops with task reductions.
3989 (expand_omp_sections): Handle expansion of sections with task
3991 (expand_omp_synch): For host teams call expand_omp_taskreg.
3992 (omp_memory_order_to_memmodel): New function.
3993 (expand_omp_atomic_load, expand_omp_atomic_store,
3994 expand_omp_atomic_fetch_op): Use it and gimple_omp_atomic_memory_order
3995 instead of gimple_omp_atomic_seq_cst_p.
3996 (build_omp_regions_1, omp_make_gimple_edges): Treat taskwait with
3997 depend clauses as a standalone directive.
3998 * omp-general.c (enum omp_requires): New variable.
3999 (omp_extract_for_data): Initialize have_reductemp member. Allow
4000 NE_EXPR even in OpenMP loops, transform them into LT_EXPR or
4001 GT_EXPR loops depending on incr sign. Formatting fixes.
4002 * omp-general.h (struct omp_for_data): Add have_reductemp member.
4003 (enum omp_requires): New enum.
4004 (omp_requires_mask): Declare.
4005 * omp-grid.c (grid_eliminate_combined_simd_part): Formatting fix.
4007 * omp-low.c (struct omp_context): Add task_reductions and
4008 task_reduction_map fields.
4009 (is_host_teams_ctx): New function.
4010 (is_taskreg_ctx): Return true also if is_host_teams_ctx.
4011 (use_pointer_for_field): Use is_global_var instead of
4012 TREE_STATIC || DECL_EXTERNAL, and apply only if not privatized
4014 (build_outer_var_ref): Ignore taskgroup outer contexts.
4015 (delete_omp_context): Release task_reductions and task_reduction_map.
4016 (scan_sharing_clauses): Don't add any fields for reduction clause on
4017 taskloop. Handle OMP_CLAUSE__REDUCTEMP_. Handle
4018 OMP_CLAUSE_{IN,TASK}_REDUCTION and OMP_CLAUSE_REDUCTION with task
4019 modifier. Don't ignore shared clauses in is_host_teams_ctx contexts.
4020 Handle OMP_CLAUSE_NONTEMPORAL.
4021 (add_taskreg_looptemp_clauses): Add OMP_CLAUSE__REDUCTEMP_ clause if
4023 (scan_omp_parallel): Add _reductemp_ clause if there are any reduction
4024 clauses with task modifier.
4025 (scan_omp_task): Handle taskwait with depend clauses.
4026 (finish_taskreg_scan): Move field corresponding to _reductemp_ clause
4027 first. Move also OMP_CLAUSE__REDUCTEMP_ clause in front if present.
4028 Handle GIMPLE_OMP_TEAMS like GIMPLE_OMP_PARALLEL.
4029 (scan_omp_for): Fix comment formatting.
4030 (scan_omp_teams): Handle host teams constructs.
4031 (check_omp_nesting_restrictions): Allow teams with no outer
4032 OpenMP context. Adjust diagnostics for teams strictly nested into
4033 some explicit OpenMP construct other than target. Allow OpenMP atomics
4034 inside of simd regions.
4035 (scan_omp_1_stmt): Call scan_sharing_clauses for taskgroups.
4036 (scan_omp_1_stmt) <case GIMPLE_OMP_TEAMS>: Temporarily bump
4037 taskreg_nesting_level while scanning host teams construct.
4038 (task_reduction_read): New function.
4039 (lower_rec_input_clauses): Handle OMP_CLAUSE_REDUCTION on taskloop
4040 construct. Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE__REDUCTEMP_
4041 clauses. Handle OMP_CLAUSE_REDUCTION with task modifier. Remove
4042 second argument create_tmp_var if it is NULL. Don't ignore shared
4043 clauses in is_host_teams_ctx contexts. Handle
4044 OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE on OMP_CLAUSE_FIRSTPRIVATE
4046 (lower_reduction_clauses): Ignore reduction clauses with task
4047 modifier. Remove second argument create_tmp_var if it is NULL.
4048 Initialize OMP_ATOMIC_MEMORY_ORDER to relaxed.
4049 (lower_send_clauses): Ignore reduction clauses with task modifier.
4050 Handle OMP_CLAUSE__REDUCTEMP_. Don't send anything for
4051 OMP_CLAUSE_REDUCTION on taskloop. Handle OMP_CLAUSE_IN_REDUCTION.
4052 (maybe_add_implicit_barrier_cancel): Add OMP_RETURN argument, don't
4053 rely that it is the last stmt in body so far. Ignore outer taskgroup
4055 (omp_task_reductions_find_first, omp_task_reduction_iterate,
4056 lower_omp_task_reductions): New functions.
4057 (lower_omp_sections): Handle reduction clauses with taskgroup
4058 modifiers. Adjust maybe_add_implicit_barrier_cancel caller.
4059 (lower_omp_single): Adjust maybe_add_implicit_barrier_cancel caller.
4060 (lower_omp_for): Likewise. Handle reduction clauses with taskgroup
4062 (lower_omp_taskgroup): Handle taskgroup reductions.
4063 (create_task_copyfn): Copy over OMP_CLAUSE__REDUCTEMP_ pointer.
4064 Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE_REDUCTION clauses.
4065 (lower_depend_clauses): If there are any
4066 OMP_CLAUSE_DEPEND_DEPOBJ or OMP_CLAUSE_DEPEND_MUTEXINOUTSET
4067 depend clauses, use a new array format. If OMP_CLAUSE_DEPEND_LAST is
4068 seen, assume lowering is done already and return early. Set kind
4069 on artificial depend clause to OMP_CLAUSE_DEPEND_LAST.
4070 (lower_omp_taskreg): Handle reduction clauses with task modifier on
4071 parallel construct. Handle reduction clause on taskloop construct.
4072 Handle taskwait with depend clauses.
4073 (lower_omp_1): Use lower_omp_taskreg instead of lower_omp_teams
4074 for host teams constructs.
4075 * tree.c (omp_clause_num_ops): Add in_reduction, task_reduction,
4076 nontemporal and _reductemp_ clause entries.
4077 (omp_clause_code_name): Likewise.
4078 (walk_tree_1): Handle OMP_CLAUSE_{IN,TASK}_REDUCTION,
4079 OMP_CLAUSE_NONTEMPORAL and OMP_CLAUSE__REDUCTEMP_.
4080 * tree-core.h (enum omp_clause_code): Add
4081 OMP_CLAUSE_{{IN,TASK}_REDUCTION,NONTEMPORAL,_REDUCTEMP_}.
4082 (enum omp_clause_defaultmap_kind, enum omp_memory_order): New.
4083 (struct tree_base): Add omp_atomic_memory_order field into union.
4084 Remove OMP_ATOMIC_SEQ_CST comment.
4085 (enum omp_clause_depend_kind): Add OMP_CLAUSE_DEPEND_MUTEXINOUTSET
4086 and OMP_CLAUSE_DEPEND_DEPOBJ.
4087 (struct tree_omp_clause): Add subcode.defaultmap_kind.
4088 * tree.def (OMP_TASKGROUP): Add another operand, move next to other
4089 OpenMP constructs with body and clauses operands.
4090 * tree.h (OMP_BODY): Use OMP_MASTER instead of OMP_TASKGROUP.
4091 (OMP_CLAUSES): Use OMP_TASKGROUP instead of OMP_SINGLE.
4092 (OMP_TASKGROUP_CLAUSES): Define.
4093 (OMP_CLAUSE_DECL): Use OMP_CLAUSE__REDUCTEMP_ instead of
4094 OMP_CLAUSE__LOOPTEMP_.
4095 (OMP_ATOMIC_SEQ_CST): Remove.
4096 (OMP_ATOMIC_MEMORY_ORDER, OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE,
4097 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL): Define.
4098 (OMP_CLAUSE_REDUCTION_CODE, OMP_CLAUSE_REDUCTION_INIT,
4099 OMP_CLAUSE_REDUCTION_MERGE, OMP_CLAUSE_REDUCTION_PLACEHOLDER,
4100 OMP_CLAUSE_REDUCTION_DECL_PLACEHOLDER,
4101 OMP_CLAUSE_REDUCTION_OMP_ORIG_REF): Handle
4102 OMP_CLAUSE_{,IN_,TASK_}REDUCTION.
4103 (OMP_CLAUSE_REDUCTION_TASK, OMP_CLAUSE_REDUCTION_INSCAN,
4104 OMP_CLAUSE_DEFAULTMAP_KIND, OMP_CLAUSE_DEFAULTMAP_CATEGORY,
4105 OMP_CLAUSE_DEFAULTMAP_BEHAVIOR, OMP_CLAUSE_DEFAULTMAP_SET_KIND):
4107 * tree-inline.c (remap_gimple_stmt): Remap taskgroup clauses.
4108 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
4109 OMP_CLAUSE__REDUCTEMP_, OMP_CLAUSE_NONTEMPORAL.
4110 (convert_local_omp_clauses): Likewise. Remove useless test.
4111 * tree-parloops.c (create_call_for_reduction_1): Pass
4112 OMP_MEMORY_ORDER_RELAXED as new argument to
4113 dump_gimple_omp_atomic_load and dump_gimple_omp_atomic_store.
4114 * tree-pretty-print.c (dump_omp_iterators): New function.
4115 (dump_omp_clause): Handle OMP_CLAUSE__REDUCTEMP_,
4116 OMP_CLAUSE_NONTEMPORAL, OMP_CLAUSE_{TASK,IN}_REDUCTION. Print
4117 reduction modifiers. Handle OMP_CLAUSE_DEPEND_DEPOBJ and
4118 OMP_CLAUSE_DEPEND_MUTEXINOUTSET. Print iterators in depend clauses.
4119 Print __internal__ for OMP_CLAUSE_DEPEND_LAST. Handle cancel and
4120 simd OMP_CLAUSE_IF_MODIFIERs. Handle new kinds of
4121 OMP_CLAUSE_DEFAULTMAP. Print conditional: for
4122 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL.
4123 (dump_omp_atomic_memory_order): New function.
4124 (dump_generic_node): Use it. Print taskgroup clauses. Print
4125 taskwait with depend clauses.
4126 * tree-pretty-print.h (dump_omp_atomic_memory_order): Declare.
4127 * tree-streamer-in.c (unpack_ts_omp_clause_value_fields):
4128 Handle OMP_CLAUSE_{TASK,IN}_REDUCTION.
4129 * tree-streamer-out.c (pack_ts_omp_clause_value_fields,
4130 write_ts_omp_clause_tree_pointers): Likewise.
4132 2018-11-08 David Malcolm <dmalcolm@redhat.com>
4135 * doc/invoke.texi (-fdump-ipa-): Document the "-optimized",
4136 "-missed", "-note", and "-all" sub-options.
4137 * ipa-inline.c (caller_growth_limits): Port from fprintf to dump
4139 (can_early_inline_edge_p): Likewise.
4140 (want_early_inline_function_p): Likewise.
4141 (want_inline_self_recursive_call_p): Likewise.
4142 (recursive_inlining): Likewise.
4143 (inline_small_functions): Likewise.
4144 (flatten_function): Likewise.
4145 (ipa_inline): Likewise.
4146 (inline_always_inline_functions): Likewise.
4147 (early_inline_small_functions): Likewise.
4148 (early_inliner): Likewise.
4149 * tree-inline.c (expand_call_inline): Likewise.
4151 2018-11-08 David Malcolm <dmalcolm@redhat.com>
4153 * pretty-print.c (pp_format): Handle %f.
4154 (selftest::test_pp_format): Add test of %f.
4155 * pretty-print.h (pp_double): New macro.
4157 2018-11-08 David Malcolm <dmalcolm@redhat.com>
4159 * dump-context.h (ASSERT_IS_CGRAPH_NODE): New macro.
4160 * dumpfile.c (make_item_for_dump_cgraph_node): Move to before...
4161 (dump_pretty_printer::decode_format): Implement "%C" for
4163 (selftest::test_capture_of_dump_calls): Rename "where" to
4164 "stmt_loc". Convert test_decl to a function decl and set its
4165 location. Add a symbol_table_test RAII instance and a
4166 cgraph_node, using it to test "%C" and dump_symtab_node.
4168 2018-11-08 Eric Botcazou <ebotcazou@adacore.com>
4171 * cgraphclones.c (duplicate_thunk_for_node): Also set DECL_IGNORED_P.
4173 2018-11-08 David Malcolm <dmalcolm@redhat.com>
4175 * cgraph.c: Include "selftest.h".
4176 (saved_symtab): New variable.
4177 (selftest::symbol_table_test::symbol_table_test): New ctor.
4178 (selftest::symbol_table_test::~symbol_table_test): New dtor.
4179 (selftest::test_symbol_table_test): New test.
4180 (selftest::cgraph_c_tests): New.
4181 * cgraph.h (saved_symtab): New decl.
4182 (selftest::symbol_table_test): New class.
4183 * selftest-run-tests.c (selftest::run_tests): Call
4184 selftest::cgraph_c_tests.
4185 * selftest.h (selftest::cgraph_c_tests): New decl.
4187 2018-11-08 Richard Biener <rguenther@suse.de>
4189 * tree-data-ref.h (lambda_int): New typedef.
4190 (lambda_vector_gcd): Adjust.
4191 (lambda_vector_new): Likewise.
4192 (lambda_matrix_new): Likewise.
4193 * tree-data-ref.c (print_lambda_vector): Adjust.
4195 2018-11-08 Richard Biener <rguenther@suse.de>
4197 PR tree-optimization/87929
4198 * tree-complex.c (expand_complex_comparison): Clean EH.
4200 2018-11-08 Martin Liska <mliska@suse.cz>
4202 * doc/extend.texi: Reword.
4203 * predict.c (expr_expected_value_1): Likewise.
4205 2018-11-08 Richard Biener <rguenther@suse.de>
4207 PR tree-optimization/87913
4208 * tree-ssa-phiopt.c (minmax_replacement): Turn EQ/NE compares
4209 of extreme values to ordered comparisons.
4211 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
4214 * doc/invoke.texi (Code Gen Options): Clarify -fno-common behavior.
4216 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
4218 * doc/invoke.texi: Remove leading dash from @opindex entries
4219 throughout the file.
4221 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
4224 * doc/invoke.texi (Option Summary): Add -e and --entry.
4225 (Link Options): Likewise.
4227 2018-11-07 Nathan Sidwell <nathan@acm.org>
4230 * Makefile.in (bitmap.o-warn): Add -Wno-error to unbreak
4231 --disable-checking bootstrap.
4233 2018-11-07 Hafiz Abid Qadeer <abidh@codesourcery.com>
4235 * configure: Regenerated.
4237 2018-11-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
4240 * stor-layout.c (compute_record_mode): Set TYPE_MODE of UNION_TYPE
4241 to the mode of the widest field iff the widest field has mode class
4242 MODE_INT, or MODE_PARTIAL_INT and the union would be passed by
4245 2018-11-07 Nikolai Merinov <n.merinov@inango-systems.com>
4247 * common.opt: Add -Wattribute-warning.
4248 * doc/invoke.texi: Add documentation for -Wno-attribute-warning.
4249 * expr.c (expand_expr_real_1): Add new attribute to warning_at
4250 call to allow user configure behavior of "warning" attribute.
4252 2018-11-07 Segher Boessenkool <segher@kernel.crashing.org>
4254 * target.def: Put @: after every vs., e.g., and i.e. where it is
4255 followed by whitespace.
4256 * doc/extend.texi: Ditto.
4257 * doc/fragments.texi: Ditto.
4258 * doc/gimple.texi: Ditto.
4259 * doc/implement-c.texi: Ditto.
4260 * doc/install.texi: Ditto.
4261 * doc/invoke.texi: Ditto.
4262 * doc/md.texi: Ditto.
4263 * doc/plugins.texi: Ditto.
4264 * doc/rtl.texi: Ditto.
4265 * doc/sourcebuild.texi: Ditto.
4266 * doc/tm.texi.in: Ditto.
4267 * doc/ux.texi: Ditto.
4268 * doc/tm.texi: Regenerate.
4270 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4272 * config/arm/arm-cpus.in (ares): New entry.
4273 * config/arm/arm-tables.opt: Regenerate.
4274 * config/arm/arm-tune.md: Likewise.
4275 * doc/invoke.texi (ARM Options): Document ares.
4277 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4279 * config/aarch64/aarch64-cores.def (ares): Define.
4280 * config/aarch64/aarch64-tune.md: Regenerate.
4281 * doc/invoke.texi (AArch64 Options): Document ares value for mtune.
4283 2018-11-07 Jan Hubicka <jh@suse.cz>
4285 * tree.c (fld_incomplete_type_of): Clear TREE_ADDRESSABLE.
4286 (free_lang_data_in_decl): Set TREE_ADDRESSABLE for public vars and
4287 functions; clear TYPE_DECL_SUPPRESS_DEBUG and DECL_MODE for
4290 2018-11-07 Richard Biener <rguenther@suse.de>
4292 PR tree-optimization/87914
4293 * tree-vect-loop.c (vect_is_simple_reduction): Improve detection
4295 (vectorizable_reduction): Handle shifts and rotates by dispatching
4296 to vectorizable_shift.
4297 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Handle
4298 in-loop uses of vect_nested_cycle defs. Merge cycle and internal
4300 (vectorizable_shift): Export and handle being called as
4302 (vect_analyze_stmt): Call vectorizable_shift after
4303 vectorizable_reduction.
4304 * tree-vectorizer.h (vectorizable_shift): Declare.
4306 2018-11-07 Jan Hubicka <jh@suse.cz>
4308 * ipa-devirt.c (odr_types_equivalent_p): Expect constants
4309 than const decls in TREE_VALUE of enum.
4310 (dump_type_inheritance_graph): Improve duplicate dumping.
4311 (free_enum_values): New.
4312 (build_type_inheritance_graph): Use it.
4313 * tree.c (free_lang_data_in_type): Free TYPE_VALUES of enums
4314 which are not main variants or not ODR types.
4315 (verify_type_variant): Expect variants to have no TYPE_VALUES.
4317 2018-11-07 Richard Biener <rguenther@suse.de>
4319 * ipa-inline.c (want_inline_small_function_p): Compute
4320 big_speedup_p lazily and last.
4322 2018-11-07 Jan Hubicka <jh@suse.cz>
4324 * tree.c (fld_type_variant_equal_p): Skip TYPE_ALIGN check when
4325 building incomplete variant of complete type.
4326 (fld_type_variant): Do not copy TYPE_ALIGN when building incomplete
4327 variant of complete type.
4329 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4331 * config/mips/mips.c: Fix typo in documentation of
4332 mips_loongson_ext2_prefetch_cookie.
4333 (mips_option_override): fix brain twister logical.
4334 * config/mips/mips.h: Fix typo in documentation of
4335 ISA_HAS_CTZ_CTO and define pattern.
4336 * config/mips/mips.md (prefetch): Hoist EXT2 above
4338 (prefetch_indexed): Hoist EXT2 above the EXT block.
4340 2018-11-07 Jan Hubicka <jh@suse.cz>
4342 * tree.c (free_lang_data_in_type): Add fld parameter; simplify
4343 return and parameter types of function and method types.
4344 (free_lang_data_in_cgraph): Update.
4346 2018-11-07 Martin Liska <mliska@suse.cz>
4348 PR rtl-optimization/87868
4349 * postreload-gcse.c (eliminate_partially_redundant_load): Set
4350 threshold to max_count if we would overflow.
4351 * profile-count.h: Make max_count a public constant.
4353 2018-11-07 Martin Liska <mliska@suse.cz>
4355 * mem-stats.h: Fix GNU coding style.
4357 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4359 * config/mips/gs264e.md: New.
4360 * config/mips/mips-cpus.def: Define gs264e.
4361 * config/mips/mips-tables.opt: Regenerate.
4362 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
4364 (mips_issue_rate): Add support for gs264e.
4365 (mips_multipass_dfa_lookahead): Likewise.
4366 * config/mips/mips.h: Define TARGET_GS264E and TUNE_GS264E.
4367 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs264e.
4368 (MIPS_ASE_MSA_SPEC): New.
4369 (BASE_DRIVER_SELF_SPECS): march=gs264e implies -mmsa.
4370 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS264E.
4371 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS264E.
4372 * config/mips/mips.md: Include gs264e.md.
4373 (processor): Add gs264e.
4374 * config/mips/mips.opt (MSA): Use Mask instead of Var.
4375 * doc/invoke.texi: Add gs264e to supported architectures.
4377 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4379 * config/mips/gs464e.md: New.
4380 * config/mips/mips-cpus.def: Define gs464e.
4381 * config/mips/mips-tables.opt: Regenerate.
4382 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
4384 (mips_issue_rate): Add support for gs464e.
4385 (mips_multipass_dfa_lookahead): Likewise.
4386 (mips_option_override): Enable MMI, EXT and EXT2 for gs464e.
4387 * config/mips/mips.h: Define TARGET_GS464E and TUNE_GS464E.
4388 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464e.
4389 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS464E.
4390 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS464E.
4391 * config/mips/mips.md: Include gs464e.md.
4392 (processor): Add gs464e.
4393 * doc/invoke.texi: Add gs464e to supported architectures.
4395 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4397 * config/mips/loongson3a.md: Rename to ...
4398 * config/mips/gs464.md: ... here.
4399 * config/mips/mips-cpus.def: Define gs464; Add loongson3a
4400 as an alias of gs464 processor.
4401 * config/mips/mips-tables.opt: Regenerate.
4402 * config/mips/mips.c (mips_issue_rate): Use PROCESSOR_GS464
4403 instead of PROCESSOR_LOONGSON_3A.
4404 (mips_multipass_dfa_lookahead): Use TUNE_GS464 instead of
4406 (mips_option_override): Enable MMI and EXT for gs464.
4407 * config/mips/mips.h: Rename TARGET_LOONGSON_3A to TARGET_GS464;
4408 Rename TUNE_LOONGSON_3A to TUNE_GS464.
4409 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464.
4410 (ISA_HAS_ODD_SPREG, ISA_AVOID_DIV_HILO, ISA_HAS_FUSED_MADD4,
4411 ISA_HAS_UNFUSED_MADD4): Use TARGET_GS464 instead of
4413 * config/mips/mips.md: Include gs464.md instead of loongson3a.md.
4414 (processor): Add gs464;
4415 * doc/invoke.texi: Add gs464 to supported architectures.
4417 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4419 * config/mips/mips-protos.h
4420 (mips_loongson_ext2_prefetch_cookie): New prototype.
4421 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): New.
4422 (mips_option_override): Enable TARGET_LOONGSON_EXT when
4423 TARGET_LOONGSON_EXT2 is true.
4424 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define
4425 __mips_loongson_ext2, __mips_loongson_ext_rev=2.
4426 (ISA_HAS_CTZ_CTO): New, true if TARGET_LOONGSON_EXT2.
4427 (ISA_HAS_PREFETCH): Include TARGET_LOONGSON_EXT and
4428 TARGET_LOONGSON_EXT2.
4429 (ASM_SPEC): Add mloongson-ext2 and mno-loongson-ext2.
4430 (define_insn "ctz<mode>2"): New insn pattern.
4431 (define_insn "prefetch"): Include TARGET_LOONGSON_EXT2.
4432 (define_insn "prefetch_indexed_<mode>"): Include
4433 TARGET_LOONGSON_EXT and TARGET_LOONGSON_EXT2.
4434 * config/mips/mips.opt (-mloongson-ext2): Add option.
4435 * gcc/doc/invoke.texi (-mloongson-ext2): Document.
4437 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4439 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Add
4440 __mips_loongson_ext.
4441 (MIPS_ASE_LOONGSON_EXT_SPEC): New.
4442 (BASE_DRIVER_SELF_SPECS): march=loongson3a implies
4444 (ASM_SPEC): Add mloongson-ext and mno-loongson-ext.
4445 * config/mips/mips.md (mul<mode>3, mul<mode>3_mul3_nohilo,
4446 <u>div<mode>3, <u>mod<mode>3, prefetch): Use TARGET_LOONGSON_EXT
4447 instead of TARGET_LOONGSON_3A.
4448 * config/mips/mips.opt (-mloongson-ext): Add option.
4449 * gcc/doc/invoke.texi (-mloongson-ext): Document.
4451 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4453 * config.gcc (extra_headers): Add loongson-mmiintrin.h.
4454 * config/mips/loongson.md: Move to ...
4455 * config/mips/loongson-mmi.md: here; Adjustment.
4456 * config/mips/loongson.h: Move to ...
4457 State as deprecated. Include loongson-mmiintrin.h for back
4458 compatibility and warning.
4459 * config/mips/loongson-mmiintrin.h: ... here.
4460 * config/mips/mips.c (mips_hard_regno_mode_ok_uncached,
4461 mips_vector_mode_supported_p, AVAIL_NON_MIPS16): Use
4462 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
4463 (mips_option_override): Make sure MMI use hard float;
4464 (mips_shift_truncation_mask, mips_expand_vpc_loongson_even_odd,
4465 mips_expand_vpc_loongson_pshufh, mips_expand_vpc_loongson_bcast,
4466 mips_expand_vector_init): Use TARGET_LOONGSON_MMI instead of
4467 TARGET_LOONGSON_VECTORS.
4468 * gcc/config/mips/mips.h (TARGET_LOONGSON_VECTORS): Delete.
4469 (TARGET_CPU_CPP_BUILTINS): Add __mips_loongson_mmi.
4470 (MIPS_ASE_DSP_SPEC, MIPS_ASE_LOONGSON_MMI_SPEC): New.
4471 (BASE_DRIVER_SELF_SPECS): march=loongson2e/2f/3a implies
4473 (SHIFT_COUNT_TRUNCATED): Use TARGET_LOONGSON_MMI instead of
4474 TARGET_LOONGSON_VECTORS.
4475 * gcc/config/mips/mips.md (MOVE64, MOVE128): Use
4476 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
4477 (Loongson MMI patterns): Include loongson-mmi.md instead of
4479 * gcc/config/mips/mips.opt (-mloongson-mmi): New option.
4480 * gcc/doc/invoke.texi (-mloongson-mmi): Document.
4482 2018-11-07 Richard Biener <rguenther@suse.de>
4485 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Fixup
4486 BLOCK_ABSTRACT_ORIGIN to be the ultimate origin.
4488 2018-11-07 Alexandre Oliva <aoliva@redhat.com>
4490 PR rtl-optimization/87874
4491 * lra.c (lra_substitute_pseudo): Do not create a subreg for
4494 2018-11-06 Aaron Sawdey <acsawdey@linux.ibm.com>
4496 * config/rs6000/rs6000.md (bswap<mode>2): Force address into register
4497 if not in indexed or indirect form.
4498 (bswap<mode>2_load): Change predicate to indexed_or_indirect_operand.
4499 (bswap<mode>2_store): Ditto.
4501 2018-11-06 Richard Earnshaw <rearnsha@arm.com>
4503 * config/aarch64/aarch64.md (speculation_tracker): Set the mode for
4506 2018-11-06 Richard Biener <rguenther@suse.de>
4508 PR tree-optimization/86850
4509 * vec.h (vec<T, va_heap, vl_ptr>::splice): Check src.length ()
4510 instead of src.m_vec.
4512 2018-11-06 Jan Hubicka <jh@suse.cz>
4514 * tree.c (fld_simplified_type_name): Break out form ...
4515 (free_lang_data_in_type): ... here.
4516 (fld_type_variant_equal_p): Use it.
4518 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4520 * config/default-d.c: Include memmodel.h.
4522 * config/sol2-d.c: New file.
4523 * config/t-sol2 (sol2-d.o): New rule.
4524 * config.gcc <*-*-solaris2*>: Set d_target_objs,
4525 target_has_targetdm.
4527 2018-11-06 Jan Hubicka <jh@suse.cz>
4529 * tree.c (fld_type_variant): Also copy alignment; be sure that
4530 new variant is equal.
4532 2018-11-06 Ilya Leoshkevich <iii@linux.ibm.com>
4535 * config/s390/s390.md: Add relative_long attribute.
4537 2018-11-06 Jan Hubicka <jh@suse.cz>
4539 * ipa-pure-const.c (check_decl): Do not test TYPE_NEEDS_CONSTRUCTING.
4540 * lto-streamer-out.c (hash_tree): Do not hash TYPE_NEEDS_CONSTRUCTING.
4541 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Do not
4542 stream TYPE_NEEDS_CONSTRUCTING.
4543 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
4544 * tree.c (free_lang_data_in_type): Clear TYPE_NEEDS_CONSTRUCTING.
4546 2018-11-06 Richard Biener <rguenther@suse.de>
4548 * tree-vect-slp.c (vect_slp_bb): Move opening of vect_slp_analyze_bb
4550 (vect_slp_analyze_bb_1): ... here to avoid hiding optimized locations.
4552 2018-11-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
4554 * gcc/config/msp430/msp430.h (REG_CLASS_CONTENTS): Add R0 to
4555 REG_CLASS_CONTENTS[GEN_REGS].
4556 (REGNO_REG_CLASS): Return NO_REGS for R2 and R3.
4558 2018-11-06 Jan Hubicka <jh@suse.cz>
4560 * tree.c (fld_simplified_type_of): Clear TYPELESS_STORAGE flag.
4562 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4565 * config/sol2.h (ASAN_CC1_SPEC): Define.
4566 (LD_WHOLE_ARCHIVE_OPTION): Define.
4567 (LD_NO_WHOLE_ARCHIVE_OPTION): Define.
4568 (ASAN_REJECT_SPEC): Provide default.
4569 (LIBASAN_EARLY_SPEC): Define.
4570 (LIBTSAN_EARLY_SPEC): Define.
4571 (LIBLSAN_EARLY_SPEC): Define.
4572 * config/i386/sol2.h (CC1_SPEC): Redefine.
4573 (ASAN_REJECT_SPEC): Define.
4575 * config/sparc/sparc.c (sparc_asan_shadow_offset): Declare.
4576 (TARGET_ASAN_SHADOW_OFFSET): Define.
4577 (sparc_asan_shadow_offset): New function.
4578 * config/sparc/sol2.h (CC1_SPEC): Append ASAN_CC1_SPEC.
4579 (ASAN_REJECT_SPEC): Define.
4581 2018-11-06 Jan Hubicka <jh@suse.cz>
4583 * tree.c (fld_type_variant): Copy canonical type.
4584 (fld_incomplete_type_of): Check that canonical types looks sane;
4585 copy canonical type.
4586 (verify_type): Accept when incomplete type has complete canonical type.
4588 2018-11-06 Jan Hubicka <jh@suse.cz>
4590 * tree.c (free_lang_data): Reset overwite_assembler_name,
4591 print_xnode, print_decl, print_type and print_identifier of
4594 2018-11-06 Richard Biener <rguenther@suse.de>
4596 PR tree-optimization/87889
4597 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
4598 Do nothing if old and new arg are the same
4600 2018-11-06 Andreas Krebbel <krebbel@linux.ibm.com>
4603 * config/s390/s390.md ("*r<noxa>sbg_di_rotl"): Remove mode
4604 attributes for operands 3 and 4.
4606 2018-11-06 Richard Biener <rguenther@suse.de>
4609 * simplify-rtx.c (simplify_binary_operation_1): Add pattern
4610 matching bitfield insertion.
4612 2018-11-06 Alexandre Oliva <aoliva@redhat.com>
4614 * auto-inc-dec.c: Include valtrack.h. Improve comments.
4615 (reg_next_debug_use): New.
4616 (attempt_change): Propagate adjusted expression into affected
4618 (merge_in_block): Track uses in debug insns.
4619 (pass_inc_dec::execute): Allocate and release
4622 2018-11-06 Wei Xiao <wei3.xiao@intel.com>
4624 * config/i386/avx512fintrin.h: Update VFIXUPIMM* intrinsics.
4625 (_mm512_fixupimm_round_pd): Update parameters and builtin.
4626 (_mm512_maskz_fixupimm_round_pd): Ditto.
4627 (_mm512_fixupimm_round_ps): Ditto.
4628 (_mm512_maskz_fixupimm_round_ps): Ditto.
4629 (_mm_fixupimm_round_sd): Ditto.
4630 (_mm_maskz_fixupimm_round_sd): Ditto.
4631 (_mm_fixupimm_round_ss): Ditto.
4632 (_mm_maskz_fixupimm_round_ss): Ditto.
4633 (_mm512_fixupimm_pd): Ditto.
4634 (_mm512_maskz_fixupimm_pd): Ditto.
4635 (_mm512_fixupimm_ps): Ditto.
4636 (_mm512_maskz_fixupimm_ps): Ditto.
4637 (_mm_fixupimm_sd): Ditto.
4638 (_mm_maskz_fixupimm_sd): Ditto.
4639 (_mm_fixupimm_ss): Ditto.
4640 (_mm_maskz_fixupimm_ss): Ditto.
4641 (_mm512_mask_fixupimm_round_pd): Update builtin.
4642 (_mm512_mask_fixupimm_round_ps): Ditto.
4643 (_mm_mask_fixupimm_round_sd): Ditto.
4644 (_mm_mask_fixupimm_round_ss): Ditto.
4645 (_mm512_mask_fixupimm_pd): Ditto.
4646 (_mm512_mask_fixupimm_ps): Ditto.
4647 (_mm_mask_fixupimm_sd): Ditto.
4648 (_mm_mask_fixupimm_ss): Ditto.
4649 * config/i386/avx512vlintrin.h:
4650 (_mm256_fixupimm_pd): Update parameters and builtin.
4651 (_mm256_maskz_fixupimm_pd): Ditto.
4652 (_mm256_fixupimm_ps): Ditto.
4653 (_mm256_maskz_fixupimm_ps): Ditto.
4654 (_mm_fixupimm_pd): Ditto.
4655 (_mm_maskz_fixupimm_pd): Ditto.
4656 (_mm_fixupimm_ps): Ditto.
4657 (_mm_maskz_fixupimm_ps): Ditto.
4658 (_mm256_mask_fixupimm_pd): Update builtin.
4659 (_mm256_mask_fixupimm_ps): Ditto.
4660 (_mm_mask_fixupimm_pd): Ditto.
4661 (_mm_mask_fixupimm_ps): Ditto.
4662 * config/i386/i386-builtin-types.def: Add new types and remove useless ones.
4663 * config/i386/i386-builtin.def: Update builtin definitions.
4664 * config/i386/i386.c: Handle new builtin types and remove useless ones.
4665 * config/i386/sse.md: Update VFIXUPIMM* patterns.
4666 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
4667 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
4668 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Update.
4669 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
4670 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
4671 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Update.
4672 * config/i386/subst.md:
4673 (round_saeonly_sd_mask_operand4): Add new subst_attr.
4674 (round_saeonly_sd_mask_op4): Ditto.
4675 (round_saeonly_expand_operand5): Ditto.
4676 (round_saeonly_expand): Update.
4678 2018-11-05 Max Filippov <jcmvbkbc@gmail.com>
4680 * config/xtensa/uclinux.h (XTENSA_ALWAYS_PIC): Change to 0.
4682 2018-11-05 Segher Boessenkool <segher@kernel.crashing.org>
4684 PR rtl-optimization/87871
4685 * combine.c (make_more_copies): Skip if dest is frame_pointer_rtx.
4687 2018-11-05 Paul Koning <ni1d@arrl.net>
4689 * doc/sourcebuild.texi (target attributes): Document new "inf"
4690 effective target keyword.
4692 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
4694 * config/s390/s390.c (s390_register_move_cost): Increase costs for
4695 moves involving the CC reg.
4697 2018-11-05 Richard Biener <rguenther@suse.de>
4699 PR tree-optimization/87873
4700 * tree-ssa-loop-manip.h (split_loop_exit_edge): Add copy_constants_p
4702 * tree-ssa-loop-manip.c (split_loop_exit_edge): Likewise.
4703 * tree-vect-loop.c (vect_transform_loop): When splitting the
4704 loop exit also create forwarder PHIs for constants.
4705 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
4706 Handle constant to_arg, add extra checking we match up the correct
4709 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
4711 * config/s390/s390.md: QImode and HImode for load on condition.
4713 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
4715 * config/s390/predicates.md: Fix typo.
4716 * config/s390/s390.md: Allow immediates for load on condition.
4718 2018-11-05 Martin Liska <mliska@suse.cz>
4720 * alloc-pool.h (struct pool_usage): Use SIZE_AMOUNT.
4721 * bitmap.h (struct bitmap_usage): Likewise.
4722 * ggc-common.c (SCALE): Remove.
4724 (struct ggc_usage): Use SIZE_AMOUNT. And update
4726 * ggc-page.c (SCALE): Remove.
4727 (STAT_LABEL): Remove.
4728 (ggc_print_statistics): Use SIZE_AMOUNT.
4729 * gimple.h (SCALE): Remove.
4731 * input.c (ONE_K): Remove.
4734 (STAT_LABEL): Likewise.
4735 (FORMAT_AMOUNT): Likewise.
4736 (dump_line_table_statistics): Use SIZE_AMOUNT.
4737 * mem-stats.h (struct mem_usage): Likewise.
4738 * rtl.c (dump_rtx_statistics): Likewise.
4739 (rtx_alloc_counts): Change type to size_t.
4740 (rtx_alloc_sizes): Likewise.
4741 (rtx_count_cmp): New.
4742 (dump_rtx_statistics): Sort first based on counts.
4743 * tree.c (tree_nodes_cmp): New.
4744 (tree_codes_cmp): New.
4745 (dump_tree_statistics): Sort first based on counts.
4746 * system.h (ONE_K): New.
4748 (SIZE_SCALE): Likewise.
4749 (SIZE_LABEL): Likewise.
4750 (SIZE_AMOUNT): Likewise.
4751 * tree-cfg.c (dump_cfg_stats): Use SIZE_AMOUNT.
4752 * tree-dfa.c (dump_dfa_stats): Likewise.
4753 * tree-phinodes.c (phinodes_print_statistics): Likewise.
4754 * tree-ssanames.c (ssanames_print_statistics): Likewise.
4755 * tree.c (dump_tree_statistics): Likewise.
4756 * vec.c (struct vec_usage): Likewise.
4757 * trans-mem.c (tm_mangle): Enlarge buffer in order to not
4758 trigger a -Werror=format-overflow with
4759 --enable-gather-detailed-stats.
4761 2018-11-05 Martin Liska <mliska@suse.cz>
4763 * mem-stats.h (mem_alloc_description::release_instance_overhead):
4765 * vec.c (struct vec_usage): Register m_element_size.
4766 (vec_prefix::register_overhead): New arguments: elements and
4768 (vec_prefix::release_overhead): Subtract elements.
4769 * vec.h (struct vec_prefix): Change signature.
4770 (va_heap::reserve): Pass proper arguments.
4771 (va_heap::release): Likewise.
4773 2018-11-05 Martin Liska <mliska@suse.cz>
4775 * mem-stats.h (mem_alloc_description::get_list): Fix GNU coding
4779 2018-11-05 Richard Biener <rguenther@suse.de>
4781 * tree-scalar-evolution.h (final_value_replacement_loop): Update
4783 * tree-scalar-evolution.c (final_value_replacement_loop): Return
4784 whether anything was done.
4785 (scev_const_prop): Remove constant propagation part, fold
4787 * tree-ssa-loop.c (pass_scev_cprop::execute): ... here.
4788 (pass_data_scev_cprop): TODO_cleanup_cfg is now done
4791 2018-11-05 Jakub Jelinek <jakub@redhat.com>
4793 PR tree-optimization/87859
4794 * gimple-ssa-store-merging.c (struct merged_store_group): Add
4795 only_constants and first_nonmergeable_order members.
4796 (merged_store_group::merged_store_group): Initialize them.
4797 (merged_store_group::do_merge): Clear only_constants member if
4798 adding something other than INTEGER_CST store.
4799 (imm_store_chain_info::coalesce_immediate_stores): Don't merge
4800 stores with order >= first_nonmergeable_order. Use
4801 merged_store->only_constants instead of always recomputing it.
4802 Set merged_store->first_nonmergeable_order if we've skipped any
4803 stores. Attempt to merge overlapping INTEGER_CST stores that
4804 we would otherwise skip.
4807 * match.pd (X + Y < X): Don't optimize if TYPE_OVERFLOW_SANITIZED.
4809 2018-11-05 Xuepeng Guo <xuepeng.guo@intel.com>
4812 * config/i386/emmintrin.h (__v16qs): New to cope with option
4814 (_mm_cmpeq_epi8): Replace __v16qi with __v16qs.
4815 (_mm_cmplt_epi8): Likewise.
4816 (_mm_cmpgt_epi8): Likewise.
4818 2018-11-05 Richard Biener <rguenther@suse.de>
4820 PR rtl-optimization/87852
4821 * fwprop.c (use_killed_between): Only consider single-defs of the
4822 use whose definition statement dominates the use.
4824 2018-11-05 Martin Liska <mliska@suse.cz>
4827 * doc/invoke.texi: Remove options that are
4828 not disabled with -Os.
4830 2018-11-05 Martin Liska <mliska@suse.cz>
4833 * doc/extend.texi: Update constrain about the last argument
4834 of __builtin_expect_with_probability.
4836 2018-11-05 Martin Liska <mliska@suse.cz>
4839 * predict.c (expr_expected_value_1): Verify
4840 that last argument is a real constants and emit
4843 2018-11-05 Martin Liska <mliska@suse.cz>
4845 PR gcov-profile/77698
4846 * ipa-profile.c (ipa_profile): Adjust hotness threshold
4849 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4851 PR tree-optimization/86572
4852 * builtins.c (c_strlen): Handle negative offsets in a safe way.
4854 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4856 PR tree-optimization/87672
4857 * gimple-fold.c (gimple_fold_builtin_stxcpy_chk): Gimplify.
4858 * tree-ssa-strlen.c (handle_builtin_strcat): Adjust object size.
4860 2018-11-04 Uros Bizjak <ubizjak@gmail.com>
4863 * cfgexpand.c (pass_expand::execute): Move the call to
4864 finish_eh_generation in front of the call to expand_stack_alignment.
4866 2018-11-04 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
4868 * common/config/i386/i386-common.c (processor_alias_table): Add
4870 * config.gcc (i[34567]86-*-linux* | ...): Add znver2.
4871 (case ${target}): Add znver2.
4872 * config/i386/driver-i386.c: (host_detect_local_cpu): Let
4873 -march=native recognize znver2 processors.
4874 * config/i386/i386-c.c (ix86_target_macros_internal): Add znver2.
4875 * config/i386/i386.c (m_znver2): New definition.
4876 (m_ZNVER): New definition.
4877 (m_AMD_MULTIPLE): Includes m_znver2.
4878 (processor_cost_table): Add znver2 entry.
4879 (processor_target_table): Add znver2 entry.
4880 (get_builtin_code_for_version): Set priority for
4882 (processor_model): Add M_AMDFAM17H_ZNVER2.
4883 (arch_names_table): Ditto.
4884 (ix86_reassociation_width): Include znver2.
4885 * config/i386/i386.h (TARGET_znver2): New definition.
4886 (struct ix86_size_cost): Add TARGET_ZNVER2.
4887 (enum processor_type): Add PROCESSOR_ZNVER2.
4888 * config/i386/i386.md (define_attr "cpu"): Add znver2.
4889 * config/i386/x86-tune-costs.h: (processor_costs) Add znver2 costs.
4890 * config/i386/x86-tune-sched.c: (ix86_issue_rate): Add znver2.
4891 (ix86_adjust_cost): Add znver2.
4892 * config/i386/x86-tune.def: Replace m_ZNVER1 by m_ZNVER.
4893 * gcc/doc/extend.texi: Add details about znver2.
4894 * gcc/doc/invoke.texi: Add details about znver2.
4896 2018-11-03 Sandra Loosemore <sandra@codesourcery.com>
4900 * config/nios2/nios2.c (nios2_rtx_costs): Recognize <mul>sidi3
4903 2018-11-02 Aaron Sawdey <acsawdey@linux.ibm.com>
4905 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Pay
4906 attention to TARGET_AVOID_XFORM and BYTES_BIG_ENDIAN.
4908 2018-11-02 Richard Earnshaw <rearnsha@arm.com>
4910 * config/aarch64/aarch64.c ((aarch64_override_options): Disable
4911 shrink-wrapping when -mtrack-speculation.
4913 2018-11-02 Richard Biener <rguenther@suse.de>
4915 * tree-ssa-coalesce.c (struct coalesce_list): Add obstack member.
4916 (pop_cost_one_pair): Do not free pair.
4917 (pop_best_coalesce): Likewise.
4918 (create_coalesce_list): Initialize obstack.
4919 (delete_coalesce_list): Free obstack.
4920 (find_coalesce_pair): Obstack-allocate coalesce pairs.
4921 (add_cost_one_coalesce): Likewise.
4922 (struct live_track): Remove bitmap pointer indirections.
4923 (new_live_track): Adjust.
4924 (delete_live_track): Likewise.
4925 (live_track_remove_partition): Likewise.
4926 (live_track_add_partition): Likewise.
4927 (live_track_live_p): Likewise.
4928 (live_track_process_def): Likewise.
4929 (live_track_clear_base_vars): Likewise.
4931 2018-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4933 * configure.ac (gcc_cv_as_sparc_register_op): Remove.
4934 * configure: Regenerate.
4935 * config.in: Regenerate.
4936 * config/sparc/sparc.c (sparc_output_scratch_registers): Remove
4937 HAVE_AS_REGISTER_PSEUDO_OP guard.
4938 * config/sparc/sparc.h (ASM_ARCH64_SPEC): Remove
4939 !HAVE_AS_REGISTER_PSEUDO_OP support.
4940 (ASM_DECLARE_REGISTER_GLOBAL): Remove HAVE_AS_REGISTER_PSEUDO_OP
4943 2018-11-02 Richard Biener <rguenther@suse.de>
4945 * dwarf2out.c (build_abbrev_table): Guard lookup_external_ref call
4948 2018-11-02 Richard Biener <rguenther@suse.de>
4950 PR tree-optimization/87776
4951 * tree-ssa-sccvn.c (do_rpo_vn): Do not mark backedges
4952 executable when iterating but running into rpo-vn-max-loop-depth
4953 and not eliding the iteration.
4955 2018-11-30 Jan Hubicka <jh@suse.cz>
4957 * tree.c (free_lang_data_in_decl): Free TREE_TYPE and alignment of
4960 2018-11-01 Aaron Sawdey <acsawdey@linux.ibm.com>
4962 * config/rs6000/rs6000-protos.h (rs6000_address_for_fpconvert): Remove
4964 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem):
4965 Combine with rs6000_address_for_fpconvert.
4966 (rs6000_address_for_fpconvert): Combine with
4967 rs6000_force_indexed_or_indirect_mem.
4968 (rs6000_expand_vector_init): Change function call from
4969 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
4970 * config/rs6000/rs6000.md (floatsi<mode>2_lfiwax): Change call from
4971 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
4972 (floatsi<mode>2_lfiwax_mem): Ditto.
4973 (floatunssi<mode>2_lfiwzx): Ditto.
4974 (floatunssi<mode>2_lfiwzx_mem): Ditto.
4975 (float<QHI:mode><FP_ISA3:mode>2): Ditto.
4976 (floatuns<QHI:mode><FP_ISA3:mode>2): Ditto.
4977 (fix_trunc<mode>si2_stfiwx): Ditto.
4978 (fixuns_trunc<mode>si2_stfiwx): Ditto.
4979 (float_<mode>si2_hw): Ditto.
4980 (floatuns_<mode>si2_hw): Ditto.
4981 * config/rs6000/vsx.md (*vsx_extract_si): Ditto.
4982 (vsx_splat_<mode>): Ditto.
4984 2018-11-01 Joseph Myers <joseph@codesourcery.com>
4986 * configure.ac (gcc_cv_as_ia64_ltoffx_ldxmov_relocs): Add newline
4987 at end of assembler input text.
4988 * configure: Regenerate.
4990 2018-11-01 Jakub Jelinek <jakub@redhat.com>
4992 PR tree-optimization/87826
4993 * gimple-ssa-store-merging.c (do_shift_rotate): Punt if count is
4994 negative or larger or equal to type's precision.
4996 2018-10-31 Alexandre Oliva <aoliva@redhat.com>
4998 * opts.c (default_options_table): Do not enable
4999 OPT_fdelayed_branch at -Og.
5000 * doc/invoke.texi (-fdelayed-branch): Document it.
5002 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
5004 * optabs-libfuncs.c (build_libfunc_function_visibility):
5005 New, split out from...
5006 (build_libfunc_function): ... here.
5007 (init_one_libfunc_visibility): New, split out from ...
5008 (init_one_libfunc): ... here.
5010 * config/aarch64/atomics.md (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse):
5011 scratch register need not be early-clobber. Document the reason
5012 why we cannot use ST<OP>.
5014 2018-10-31 Joseph Myers <joseph@codesourcery.com>
5017 * configure.ac: Remove AC_PREREQ. Use AC_LANG_SOURCE. Use single
5018 line for second argument of AC_DEFINE_UNQUOTED.
5019 * doc/install.texi (Tools/packages necessary for modifying GCC):
5020 Update to autoconf 2.69 and automake 1.15.1.
5021 * aclocal.m4, config.in, configure: Regenerate.
5023 2018-10-31 Pat Haugen <pthaugen@us.ibm.com>
5025 * tree-ssa-loop-ivcanon.c (try_peel_loop): Fix typo and remove dead
5028 2018-10-31 Martin Liska <mliska@suse.cz>
5031 * common/config/aarch64/aarch64-common.c (aarch64_parse_extension):
5032 Add new argument invalid_extension.
5033 (aarch64_get_all_extension_candidates): New function.
5034 (aarch64_rewrite_selected_cpu): Add NULL to function call.
5035 * config/aarch64/aarch64-protos.h (aarch64_parse_extension): Add
5037 (aarch64_get_all_extension_candidates): New function.
5038 * config/aarch64/aarch64.c (aarch64_parse_arch): Add new
5039 argument invalid_extension.
5040 (aarch64_parse_cpu): Likewise.
5041 (aarch64_print_hint_for_extensions): New function.
5042 (aarch64_validate_mcpu): Provide hint about invalid extension.
5043 (aarch64_validate_march): Likewise.
5044 (aarch64_handle_attr_arch): Pass new argument.
5045 (aarch64_handle_attr_cpu): Provide hint about invalid extension.
5046 (aarch64_handle_attr_isa_flags): Likewise.
5048 2018-10-31 Richard Biener <rguenther@suse.de>
5052 * tree-outof-ssa.c (insert_backedge_copies): Restrict
5053 copy generation to useful cases. Place the copy before
5054 the definition of the backedge value when possible.
5056 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
5058 * config/arc/arc-arch.h (ARC_TUNE_ARC7XX): New tune value.
5059 * config/arc/arc.c (arc_active_insn): New function.
5060 (check_store_cacheline_hazard): Likewise.
5061 (workaround_arc_anomaly): Use check_store_cacheline_hazard.
5062 (arc_override_options): Disable delay slot scheduler for older
5064 (arc_store_addr_hazard_p): New implementation, old one renamed to
5066 (arc_store_addr_hazard_internal_p): Renamed.
5067 (arc_reorg): Don't combine into brcc instructions which are part
5068 of hardware hazard solution.
5069 * config/arc/arc.md (attr tune): Consider new arc7xx tune value.
5070 (tune_arc700): Likewise.
5071 * config/arc/arc.opt (arc7xx): New tune value.
5072 * config/arc/arc700.md: Improve A7 scheduler.
5074 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
5076 * config/arc/arc.c (arc_override_options): Remove
5077 TARGET_COMPACT_CASESI.
5078 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Update.
5079 (CASE_VECTOR_MODE): Likewise.
5080 (CASE_VECTOR_PC_RELATIVE): Likewise.
5081 (CASE_VECTOR_SHORTEN_MODE): Likewise.
5082 (CASE_VECTOR_SHORTEN_MODE1): Delete.
5083 (ADDR_VEC_ALIGN): Update.
5084 (ASM_OUTPUT_CASE_LABEL): Undefine.
5085 (ASM_OUTPUT_BEFORE_CASE_LABEL): Undefine.
5086 (TARGET_BI_BIH): Define.
5087 (DEFAULT_BRANCH_INDEX): Likewise.
5088 * config/arc/arc.md (casesi): Rework to accept BI/BIH
5089 instructions, remove compact_casesi use case.
5090 (casesi_compact_jump): Remove.
5091 (casesi_dispatch): New pattern.
5092 * config/arc/arc.opt: Add mbranch-index option. Deprecate
5093 compact_casesi option.
5094 * doc/invoke.texi: Document mbranch-index option.
5096 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
5098 * config/arc/arc.c (arc_get_tp): Remove function.
5099 (arc_emit_call_tls_get_addr): Likewise.
5100 (arc_call_tls_get_addr): New function.
5101 (arc_legitimize_tls_address): Make use of arc_call_tls_get_addr.
5102 * config/arc/arc.md (tls_load_tp_soft): Remove.
5103 (tls_gd_get_addr): Likewise.
5105 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
5107 * config/arc/arc.md (mulsi3): Remove call to mulsi_600_lib.
5108 (mulsi3_600_lib): Remove pattern.
5109 (umulsi3_highpart_600_lib_le): Likewise.
5110 (umulsi3_highpart): Remove call to umulsi3_highpart_600_lib_le.
5111 (umulsidi3): Remove call to umulsidi3_600_lib.
5112 (umulsidi3_600_lib): Remove pattern.
5113 (peephole2): Remove peephole using the above deprecated patterns.
5115 2018-10-31 Thomas Preud'homme <thomas.preudhomme@linaro.org>
5118 * config/arm/arm.c (arm_option_check_internal): Disable the combined
5119 use of -mslow-flash-data and -mword-relocations.
5120 (arm_option_override): Enable -mword-relocations if -fpic or -fPIC.
5121 * config/arm/arm.md (SYMBOL_REF MOVT splitter): Stop checking for
5123 * doc/invoke.texi (-mword-relocations): Mention conflict with
5125 (-mslow-flash-data): Reciprocally.
5127 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
5129 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Force
5130 16-byte modes held in GP registers to use an even regno.
5132 * config/aarch64/aarch64.c (aarch64_emit_bic): Remove.
5133 (aarch64_atomic_ldop_supported_p): Remove.
5134 (aarch64_gen_atomic_ldop): Remove.
5135 * config/aarch64/atomic.md (atomic_<atomic_optab><ALLI>):
5136 Fully expand LSE operations here.
5137 (atomic_fetch_<atomic_optab><ALLI>): Likewise.
5138 (atomic_<atomic_optab>_fetch<ALLI>): Likewise.
5139 (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse): Drop atomic_op iterator
5140 and use ATOMIC_LDOP instead; use register_operand for the input;
5141 drop the split and emit insns directly.
5142 (aarch64_atomic_fetch_<ATOMIC_LDOP><ALLI>_lse): Likewise.
5143 (aarch64_atomic_<atomic_op>_fetch<ALLI>_lse): Remove.
5144 (@aarch64_atomic_load<ATOMIC_LDOP><ALLI>): Remove.
5146 * config/aarch64/aarch64.c (aarch64_emit_atomic_swap): Remove.
5147 (aarch64_gen_atomic_ldop): Don't call it.
5148 * config/aarch64/atomics.md (atomic_exchange<ALLI>):
5149 Use aarch64_reg_or_zero.
5150 (aarch64_atomic_exchange<ALLI>): Likewise.
5151 (aarch64_atomic_exchange<ALLI>_lse): Remove split; remove & from
5152 operand 0; use aarch64_reg_or_zero for input; merge ...
5153 (@aarch64_atomic_swp<ALLI>): ... this and remove.
5155 * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): New.
5156 (aarch64_split_compare_and_swap): Use it.
5157 (aarch64_expand_compare_and_swap): Likewise. Remove convert_modes;
5158 test oldval against the proper predicate.
5159 * config/aarch64/atomics.md (@atomic_compare_and_swap<ALLI>):
5160 Use nonmemory_operand for expected.
5161 (cas_short_expected_pred): New.
5162 (@aarch64_compare_and_swap<SHORT>): Use it; use "rn" not "rI" to match.
5163 (@aarch64_compare_and_swap<GPI>): Use "rn" not "rI" for expected.
5164 * config/aarch64/predicates.md (aarch64_plushi_immediate): New.
5165 (aarch64_plushi_operand): New.
5167 * config/aarch64/aarch64.c (aarch64_expand_compare_and_swap):
5168 Force oldval into the rval register for TARGET_LSE; emit the compare
5169 during initial expansion so that it may be deleted if unused.
5170 (aarch64_gen_atomic_cas): Remove.
5171 * config/aarch64/atomics.md (@aarch64_compare_and_swap<SHORT>_lse):
5172 Change =&r to +r for operand 0; use match_dup for operand 2;
5173 remove is_weak and mod_f operands as unused. Drop the split
5175 (@aarch64_atomic_cas<SHORT>): ... this pattern's output; remove.
5176 (@aarch64_compare_and_swap<GPI>_lse): Similarly.
5177 (@aarch64_atomic_cas<GPI>): Similarly.
5179 2018-10-31 Richard Biener <rguenther@suse.de>
5181 * tree-eh.c (replace_trapping_overflow): Simplify ABS_EXPR case
5184 2018-10-31 Sameera Deshpande <sameera.deshpande@linaro.org>
5186 * config/aarch64/aarch64-cores.def (saphira): Use saphira pipeline.
5187 * config/aarch64/aarch64.md: Include saphira.md
5188 * config/aarch64/saphira.md: New file for pipeline description.
5190 2018-10-30 Martin Sebor <msebor@redhat.com>
5193 * gimple-ssa-sprintf.c (format_directive): Use %G to include
5195 (sprintf_dom_walker::compute_format_length):
5196 Avoid setting POSUNDER4K here.
5197 (get_destination_size): Handle null argument values.
5198 (get_user_idx_format): New function.
5199 (sprintf_dom_walker::handle_gimple_call): Handle all printf-like
5200 functions, including user-defined with attribute format printf.
5201 Use %G to include inlining context.
5202 Set POSUNDER4K here.
5204 2018-10-30 Jan Hubicka <jh@suse.cz>
5206 * params.def (lto-partitions): Bump from 32 to 128.
5208 2018-10-30 Jan Hubicka <jh@suse.cz>
5211 (free_lang_data_d, add_tree_to_fld_list, fld_worklist_push): Move
5213 (free_lang_data_in_type): Forward declare.
5214 (fld_type_variant_equal_p): New function.
5215 (fld_type_variant): New function
5216 (fld_incomplete_types): New hash.
5217 (fld_incomplete_type_of): New function
5218 (fld_simplfied-type): New function.
5219 (free_lang_data_in_decl): Add fld parameter; simplify type of FIELD_DECL
5220 (free_lang_data): Allocate and free fld_incomplete_type; update call
5221 of free_lang_data_in_decl.
5223 2018-10-30 Eric Botcazou <ebotcazou@adacore.com>
5225 * gcov.c (output_lines): Remove duplicate line.
5227 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
5229 * config/rs6000/rs6000.md (bswapdi2): Force address into register
5230 if not in indexed or indirect form.
5231 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
5232 (bswapdi2_store): Ditto.
5233 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
5235 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
5236 Prototype for helper function.
5238 2018-10-30 Martin Sebor <msebor@redhat.com>
5240 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
5241 (target, pragma GCC optimize, pragma GCC target): Ditto.
5243 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
5245 * doc/extend.texi: Fix prototype and description of
5246 __builtin_expect_with_probability.
5248 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
5250 * cgraph.h (clone_function_name_1): Replaced by new
5251 clone_function_name_numbered that takes name as string; for
5252 privatize_symbol_name_1 use only.
5253 (clone_function_name): Renamed to
5254 clone_function_name_numbered to be explicit about numbering.
5255 (clone_function_name): New two-argument function that does
5256 not number its output.
5257 (clone_function_name): New three-argument function that
5258 takes a number to append to its output.
5259 * cgraphclones.c (duplicate_thunk_for_node):
5260 (clone_function_name_1): Renamed.
5261 (clone_function_name_numbered): Two new functions.
5262 (clone_function_name): Improved documentation.
5263 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
5264 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
5265 * final.c (final_scan_insn_1): Use the new clone_function_name
5267 * multiple_target.c (create_dispatcher_calls): Ditto.
5268 (create_target_clone): Ditto.
5269 * omp-expand.c (grid_expand_target_grid_body): Ditto.
5270 * omp-low.c (create_omp_child_function_name): Ditto.
5271 * omp-simd-clone.c (simd_clone_create): Ditto.
5272 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
5273 new clone_function_name without numbering.
5275 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
5277 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
5278 Assert that the allocation size is not zero.
5280 2018-10-30 Richard Biener <rguenther@suse.de>
5282 PR tree-optimization/87800
5283 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
5284 non-induction or reduction PHIs.
5286 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
5288 * config/aarch64/falkor-tag-collision-avoidance.c
5289 (execute_tag_collision_avoidance): Call df_note_add_problem.
5291 2018-10-30 Martin Liska <mliska@suse.cz>
5293 * doc/extend.texi: Fix typo in documentation
5294 of __builtin_expect_with_probability.
5296 2018-10-29 David Malcolm <dmalcolm@redhat.com>
5299 * input.c (get_substring_ranges_for_loc): Detect if
5300 linemap_resolve_location gives us a NULL map, and reject
5303 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
5305 * config.gcc (xstormy16-*-elf): Set tm_d_file.
5307 2018-10-29 David Malcolm <dmalcolm@redhat.com>
5308 Martin Sebor <msebor@redhat.com>
5309 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
5311 * doc/ux.texi (Quoting): New subsection, adapted from material at
5312 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
5313 MartinSebor and ManuelLopezIbanez.
5314 (Fix-it hints): Note that fix-it hints shouldn't be marked for
5317 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
5320 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
5323 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
5325 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
5327 2018-10-29 Paul Koning <ni1d@arrl.net>
5329 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
5331 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
5333 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
5334 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
5335 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
5336 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
5337 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
5338 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
5339 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
5340 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
5341 Change 'vector' to '__vector'.
5342 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
5343 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
5344 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
5345 _mm_avg_pu16): Likewise. And, whitespace corrections.
5347 2018-10-29 Richard Biener <rguenther@suse.de>
5349 PR tree-optimization/87785
5350 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
5353 2018-10-29 Olivier Hainque <hainque@adacore.com>
5355 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
5357 2018-10-29 Olivier Hainque <hainque@adacore.com>
5359 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
5360 ports configuration macro, defaults to "gnu".
5361 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
5362 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
5364 2018-10-29 Olivier Hainque <hainque@adacore.com>
5366 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
5368 2018-10-29 Richard Biener <rguenther@suse.de>
5370 PR tree-optimization/87790
5371 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
5372 (vect_make_slp_decision): Adjust.
5373 (vect_slp_analyze_bb_1): Likewise.
5374 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
5377 2018-10-29 Richard Biener <rguenther@suse.de>
5379 PR tree-optimization/87785
5380 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
5382 (vect_build_slp_tree): Likewise.
5383 (vect_gather_slp_loads): New function.
5384 (vect_analyze_slp_instance): Gather loads separately from the
5387 2018-10-29 Martin Liska <mliska@suse.cz>
5389 * Makefile.in: Make dependency to json.o.
5390 * doc/gcov.texi: Document new JSON format, remove
5391 old intermediate format documentation.
5392 * gcov.c (struct function_info): Come up with m_name and
5394 (function_info::function_info): Initialize it.
5395 (function_info::~function_info): Release it.
5396 (main): Rename flag_intermediate_format to flag_json_format.
5397 (print_usage): Describe --json-format.
5398 (process_args): Set flag_json_format.
5399 (output_intermediate_line): Remove.
5400 (output_intermediate_json_line): Likewise.
5401 (get_gcov_intermediate_filename): Return new extension
5403 (output_intermediate_file): Implement JSON emission.
5404 (output_json_intermediate_file): Implement JSON emission.
5405 (generate_results): Use ::get_name for function name.
5406 Handle JSON output file.
5407 (read_graph_file): Use ::get_name instead of cplus_demangle.
5408 (read_count_file): Likewise.
5409 (solve_flow_graph): Likewise.
5410 (add_line_counts): Likewise.
5411 (accumulate_line_counts): Use new flag_json_format.
5412 (output_function_details): Use ::get_name instead of cplus_demangle.
5413 (output_lines): Likewise.
5414 * json.cc (test_writing_literals): Add new tests.
5415 * json.h (class literal): Add new boolean constructor.
5417 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
5419 PR rtl-optimization/87701
5420 PR rtl-optimization/87780
5421 * combine.c (make_more_copies): Rewrite.
5423 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
5425 * doc/generic.texi (ABSU_EXPR): Document.
5426 * match.pd (absu(x)*absu(x) -> x*x): Handle.
5427 (absu(absu(X)) -> absu(X)): Likewise.
5428 (absu(-X) -> absu(X)): Likewise.
5429 (absu(X) where X is nonnegative -> X): Likewise.
5431 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
5433 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
5434 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
5435 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
5436 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
5437 (s-tm-texi): Also check timestamp on d-target.def.
5438 (generated_files): Add TM_D_H and d-target-hooks-def.h.
5439 (build/genhooks.o): Also depend on D_TARGET_DEF.
5440 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
5442 * config/aarch64/aarch64-d.c: New file.
5443 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
5445 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
5447 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
5448 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
5449 * config/arm/arm-d.c: New file.
5450 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
5451 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
5452 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
5453 * config/arm/t-arm (arm-d.o): New rule.
5454 * config/default-d.c: New file.
5455 * config/glibc-d.c: New file.
5456 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5457 * config/i386/i386-d.c: New file.
5458 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
5459 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
5460 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
5461 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
5462 * config/i386/t-i386 (i386-d.o): New rule.
5463 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5464 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5465 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
5466 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5467 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
5468 * config/mips/mips-d.c: New file.
5469 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
5470 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
5471 * config/mips/t-mips (mips-d.o): New rule.
5472 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5473 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5474 * config/powerpcspe/powerpcspe-d.c: New file.
5475 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
5477 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
5478 Support GNU D by using 0 as the language type.
5479 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
5480 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
5481 * config/riscv/riscv-d.c: New file.
5482 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
5484 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
5485 * config/riscv/t-riscv (riscv-d.o): New rule.
5486 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5487 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5488 * config/rs6000/rs6000-d.c: New file.
5489 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
5491 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
5492 Support GNU D by using 0 as the language type.
5493 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
5494 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
5495 * config/s390/s390-d.c: New file.
5496 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
5497 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
5498 * config/s390/t-s390 (s390-d.o): New rule.
5499 * config/sparc/sparc-d.c: New file.
5500 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
5502 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
5503 * config/sparc/t-sparc (sparc-d.o): New rule.
5504 * config/t-glibc (glibc-d.o): New rule.
5505 * configure: Regenerated.
5506 * configure.ac (tm_d_file): New variable.
5507 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
5508 * doc/contrib.texi (Contributors): Add self for the D frontend.
5509 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
5510 * doc/install.texi (Configuration): Mention libphobos as an option for
5511 --enable-shared. Mention d as an option for --enable-languages.
5512 (Testing): Mention check-d as a target.
5513 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
5514 name suffixes. Mention d as a -x option.
5515 * doc/sourcebuild.texi (Top Level): Mention libphobos.
5516 * doc/standards.texi (Standards): Add section on D language.
5517 * doc/tm.texi: Regenerated.
5518 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
5519 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
5520 * dwarf2out.c (is_dlang): New function.
5521 (gen_compile_unit_die): Use DW_LANG_D for D.
5522 (declare_in_namespace): Return module die for D, instead of adding
5523 extra declarations into the namespace.
5524 (gen_namespace_die): Generate DW_TAG_module for D.
5525 (gen_decl_die): Handle CONST_DECLSs for D.
5526 (dwarf2out_decl): Likewise.
5527 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
5528 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
5529 kinds of aggregates.
5530 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
5531 * genhooks.c: Include d/d-target.def.
5533 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
5536 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
5537 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
5538 (STACK_DYNAMIC_OFFSET): Likewise.
5540 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
5543 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
5546 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
5548 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
5549 return type and other typos.
5551 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
5553 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
5554 a shorter sequence with fewer branches.
5555 (emit_final_str_compare_gpr): Ditto.
5557 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
5559 * config/rs6000/tmmintrin.h: New file.
5560 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
5562 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
5564 * config/rs6000/mmintrin.h: Enable 32bit compilation.
5565 * config/rs6000/xmmintrin.h: Likewise.
5567 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
5569 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
5571 2018-10-26 Richard Biener <rguenther@suse.de>
5573 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
5575 (vect_mark_slp_stmts_relevant): Likewise.
5576 (vect_detect_hybrid_slp_stmts): Likewise.
5577 (vect_bb_slp_scalar_cost): Likewise.
5578 (vect_remove_slp_scalar_calls): Likewise.
5580 2018-10-26 Jan Hubicka <jh@suse.cz>
5582 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
5583 (warn_types_mismatch): Fix walk of DECL_NAME.
5584 (odr_types_equivalent_p): Fix overactive assert.
5586 2018-10-26 Richard Biener <rguenther@suse.de>
5588 PR tree-optimization/87105
5589 * tree-vectorizer.h (_slp_tree::refcnt): New member.
5590 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
5592 (vect_create_new_slp_node): Initialize refcnt to one.
5594 (scalar_stmts_set_t, bst_fail): Remove.
5595 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
5596 (vect_build_slp_tree): Add bst_map argument and lookup
5597 already created SLP nodes.
5598 (vect_print_slp_tree): Handle a SLP graph, print SLP node
5600 (vect_slp_rearrange_stmts): Handle a SLP graph.
5601 (vect_analyze_slp_instance): Adjust and free SLP nodes from
5602 the CSE map. Fix indenting.
5603 (vect_schedule_slp_instance): Add short-cut.
5605 2018-10-26 Martin Liska <mliska@suse.cz>
5608 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
5609 addr_expr and not with pointers.
5611 2018-10-26 Jan Hubicka <jh@suse.cz>
5613 * tree.c (free_lang_data_in_type): Only check main variants.
5614 * ipa-devirt.c (warn_odr): Make static.
5615 (types_same_for_odr): Drop strict variant.
5616 (types_odr_comparable): Likewise.
5617 (odr_or_derived_type_p): Look for main variants.
5618 (odr_name_hasher::equal): Cleanup comment.
5619 (odr_subtypes_equivalent): Add warn and warned arguments; check main
5621 (type_variants_equivalent_p): break out from ...
5622 (odr_types_equivalent): ... here; go for main variants where needed.
5623 (warn_odr): ... here; turn static.
5624 (warn_types_mismatch): Compare mangled names of main variants.
5625 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
5626 (type_with_linkage_p): Sanity check that we look at main variant.
5627 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
5628 * tree.h (types_same_for_odr): Drop strict argument.
5630 2018-10-26 Richard Biener <rguenther@suse.de>
5632 PR tree-optimization/87746
5633 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
5634 Simplify and fix WRT strided store groups with size not
5635 equal to step in element count.
5636 (vect_analyze_group_access_1): Dump the whole group.
5638 2018-10-25 Carl Love <cel@us.ibm.com>
5640 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
5641 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
5642 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
5643 precicion entry for each overloaded builtin.
5644 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
5645 VSCEDPUO): Rename overloaded name.
5646 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
5647 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
5648 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
5649 define_expand for xscmpexqp instruction.
5650 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
5652 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
5653 Jinsong Ji <jji@us.ibm.com>
5655 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
5656 function with vec_sl.
5657 (_mm_slli_epi32): Likewise.
5658 (_mm_slli_epi64): Likewise.
5659 (_mm_srai_epi16): Replace deprecated function with vec_sra.
5660 (_mm_srai_epi32): Likewise.
5661 (_mm_srli_epi16): Replace deprecated function with vec_sr.
5662 (_mm_srli_epi32): Likewise.
5663 (_mm_srli_epi64): Likewise.
5664 (_mm_sll_epi16): Replace deprecated function with vec_sl.
5665 (_mm_sll_epi32): Likewise.
5666 (_mm_sll_epi64): Likewise.
5667 (_mm_sra_epi16): Replace deprecated function with vec_sra.
5668 (_mm_sra_epi32): Likewise.
5669 (_mm_srl_epi16): Replace deprecated function with vec_sr.
5670 (_mm_srl_epi32): Likewise.
5671 (_mm_srl_epi64): Likewise.
5673 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
5674 Jinsong Ji <jji@us.ibm.com>
5676 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
5677 comparison operators with vec_cmp* for compatibility due to
5678 unfortunate history; clean up formatting and use types more
5680 (_mm_sll_epi32): Likewise.
5681 (_mm_sll_epi64): Likewise.
5682 (_mm_srl_epi16): Likewise.
5683 (_mm_srl_epi32): Likewise.
5684 (_mm_srl_epi64): Likewise.
5686 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
5687 Jinsong Ji <jji@us.ibm.com>
5689 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
5690 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
5691 __vector __bool int. Use vec_cmpgt in preference to deprecated
5692 function vec_vcmpgtfp.
5693 (_mm_max_ps): Likewise.
5695 2018-10-25 Jeff Law <law@redhat.com>
5697 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
5700 2018-10-25 Martin Sebor <msebor@redhat.com>
5702 * doc/extend.texi (aligned): Expand attribute description.
5703 (Alignment): Rename section. Discuss function arguments.
5705 2018-10-25 Jan Hubicka <jh@suse.cz>
5707 * ipa-devirt.c (main_odr_variant): Remove.
5708 (hash_odr_name, types_same_for_odr, types_odr_comparable,
5709 odr_name_hasher::equal, odr_subtypes_equivalent_p):
5710 Drop use of main_odr_variant.
5711 (add_type_duplicate): Silence confused warnings on integer types.
5712 (get_odr_type): Always look for main variant.
5713 (register_odr_type): Simplify.
5715 2018-10-25 Richard Biener <rguenther@suse.de>
5717 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
5718 Initialize ng to silence error with release checking bootstrap.
5720 2018-10-25 Richard Biener <rguenther@suse.de>
5722 * tree-if-conv.c: Include tree-ssa-sccvn.h.
5723 (tree_if_conversion): Run CSE on the if-converted loop body.
5725 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
5727 * config/s390/constraints.md (ZL): New constraint.
5728 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
5730 * config/s390/s390.md (movdi_larl): Remove.
5731 (movdi_64): Add the LARL alternative.
5733 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
5736 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
5737 (RTX_CODE_HWINT_P): New macro.
5738 (rtx_code_size): Use RTX_CODE_HWINT_P ().
5740 2018-10-25 Jan Hubicka <jh@suse.cz>
5742 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
5745 2018-10-25 Richard Biener <rguenther@suse.de>
5747 PR tree-optimization/87665
5748 PR tree-optimization/87745
5749 * tree-vectorizer.h (get_earlier_stmt): Remove.
5750 (get_later_stmt): Pick up UID from the original non-pattern stmt.
5752 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
5754 * options.texi (Deprecated): Move list to Var section.
5756 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
5757 Jinsong Ji <jji@us.ibm.com>
5759 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
5760 __vector long to __vector long long.
5761 (_mm_cvtpd_ps): Likewise.
5762 (_mm_cvttpd_epi32): Likewise.
5763 (_mm_cvtpi32_pd): Likewise.
5764 (_mm_unpackhi_epi64): Likewise.
5765 (_mm_unpacklo_epi64): Likewise.
5767 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
5769 PR rtl-optimization/87720
5770 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
5772 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
5774 * gimple-ssa-isolate-paths.c
5775 (find_implicit_erroneous_behavior): Do not change code if the
5776 pass is running for warnings only.
5777 (find_explicit_erroneous_behavior): Likewise.
5779 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
5781 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
5782 Define as rs6000_mangle_decl_assembler_name.
5783 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
5784 long double to IEEE long double, switch the names of the long
5785 double built-in functions to be <func>f128 instead of <func>l.
5787 2018-10-24 Martin Sebor <msebor@redhat.com>
5789 * doc/extend.texi (nonnull): List no-argument form. Reference
5790 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
5792 2018-10-24 Richard Biener <rguenther@suse.de>
5794 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
5796 2018-10-24 Martin Liska <mliska@suse.cz>
5798 PR tree-optimization/84436
5799 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
5801 (switch_conversion::contains_linear_function_p): New.
5802 (switch_conversion::build_one_array): Support linear
5803 transformation on input.
5804 * tree-switch-conversion.h (struct switch_conversion): Add
5805 contains_linear_function_p declaration.
5807 2018-10-24 Richard Biener <rguenther@suse.de>
5809 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
5810 if its argument is CONSTANT_CLASS_P.
5812 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
5814 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
5815 it is wrong for forward declarations.
5817 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
5819 * config/s390/s390.c (s390_check_qrst_address): Add the missing
5820 SYMBOL_REF_P () check.
5822 2018-10-24 Richard Biener <rguenther@suse.de>
5824 PR tree-optimization/87105
5825 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
5826 dump classification.
5827 (vect_analyze_data_ref_accesses): Handle duplicate loads and
5828 stores by splitting the affected group after the fact.
5829 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
5830 fail the SLP build because of size constraints.
5832 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5834 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
5835 * configure: Regenerate.
5836 * config.in: Regenerate.
5837 * varasm.c (mergeable_string_section): Use readonly_data_section
5838 if linker doesn't support SHF_MERGE with alignment > 8.
5839 (mergeable_constant_section): Likewise.
5841 2018-10-24 Richard Biener <rguenther@suse.de>
5843 PR tree-optimization/84013
5844 * tree-ssa-structalias.c (struct msdi_data): New struct for
5845 marshalling data to walk_stmt_load_store_ops.
5846 (maybe_set_dependence_info): Refactor as callback for
5847 walk_stmt_load_store_ops.
5848 (compute_dependence_clique): Set restrict info on all stmt kinds.
5850 2018-10-24 Martin Liska <mliska@suse.cz>
5852 * cgraph.c (cgraph_node::dump):
5853 Remove reduntant dumps and make tp_first_run dump more compact.
5855 2018-10-24 Richard Biener <rguenther@suse.de>
5857 PR tree-optimization/87665
5858 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
5861 2018-10-12 Jeff Law <law@redhat.com>
5863 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
5866 2018-10-23 Richard Biener <rguenther@suse.de>
5868 * tree-vrp.c (add_assert_info): Guard dump_printf with
5870 * gimple-ssa-evrp-analyze.c
5871 (evrp_range_analyzer::record_ranges_from_incoming_edge):
5872 Use value_range::ignore_equivs_equal_p.
5874 2018-10-23 Richard Biener <rguenther@suse.de>
5876 PR tree-optimization/87105
5877 PR tree-optimization/87608
5878 * passes.def (pass_all_early_optimizations): Add early phi-opt
5880 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
5881 addition to debug stmts.
5882 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
5883 and abs replacement early.
5884 * tree-cfg.c (gimple_empty_block_p): Likewise.
5886 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
5889 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
5890 specified to configure.
5891 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
5893 2018-10-23 Richard Biener <rguenther@suse.de>
5895 PR tree-optimization/87700
5896 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
5898 2018-10-23 Jakub Jelinek <jakub@redhat.com>
5901 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
5902 second argument from __mmask16 to __mmask8.
5903 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
5904 _mm_mask_packs_epi32): Likewise.
5905 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
5907 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
5909 2018-10-23 Richard Biener <rguenther@suse.de>
5911 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
5913 2018-10-23 Richard Biener <rguenther@suse.de>
5915 PR tree-optimization/86144
5916 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
5917 over simd attribute.
5919 2018-10-23 Richard Biener <rguenther@suse.de>
5921 PR tree-optimization/87693
5922 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
5923 the case we do not find the taken edge.
5925 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
5926 Jinsong Ji <jji@us.ibm.com>
5928 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
5929 (_mm_store_pd): Use unaligned vector type for pointer cast.
5930 (_mm_maskmoveu_si128): Likewise.
5931 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
5932 (_mm_store_ps): Use unaligned vector type for pointer cast.
5934 2018-10-22 Paul Koning <ni1d@arrl.net>
5936 * symtab.c (symtab_node::increase_alignment): Correct max
5939 2018-10-22 Yury Gribov <tetra2005@gmail.com>
5941 PR tree-optimization/87633
5942 * match.pd: Do not generate unordered integer comparisons.
5944 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
5946 PR rtl-optimization/87600
5947 * combine.c: Add include of expr.h.
5948 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
5949 register to a pseudo.
5950 (make_more_copies): New function, add a copy to a new pseudo after
5951 the moves from hard registers into pseudos.
5952 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
5953 later. Call make_more_copies.
5955 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
5957 * lra-constraints.c (process_alt_operands): New local array,
5958 matching_early_clobber. Check matching_early_clobber before
5959 decrementing reject, and set matching_early_clobber after.
5961 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
5964 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
5965 call output_addr_const and hope for the best.
5967 2018-10-22 Richard Biener <rguenther@suse.de>
5969 * gimple-ssa-evrp-analyze.c
5970 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
5971 smarter about what ranges to use.
5972 * tree-vrp.c (add_assert_info): Dump here.
5973 (register_edge_assert_for_2): Instead of here at multiple but
5976 * gcc.dg/tree-ssa/evrp12.c: New testcase.
5977 * gcc.dg/predict-6.c: Adjust.
5978 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
5979 * gcc.dg/tree-ssa/vrp02.c: Likewise.
5980 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
5982 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
5983 Richard Biener <rguenther@suse.de>
5985 * bitmap.h: Update data structure documentation, including a
5986 description of bitmap views as either linked-lists or splay trees.
5987 (struct bitmap_element_def): Update comments for splay tree bitmaps.
5988 (struct bitmap_head_def): Likewise.
5989 (bitmap_list_view, bitmap_tree_view): New prototypes.
5990 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
5992 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
5993 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
5994 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
5995 released bitmap element here.
5996 (bitmap_element_free): Remove.
5997 (bitmap_elt_clear_from): Work on splay tree bitmaps.
5998 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
5999 this function similar ones such that linked-list bitmap implementation
6000 functions are grouped.
6001 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
6002 and moved for grouping.
6003 (bitmap_list_insert_element_after): Renamed from
6004 bitmap_elt_insert_after, and moved for grouping.
6005 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
6006 (bitmap_tree_link_left, bitmap_tree_link_right,
6007 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
6008 bitmap_tree_link_element, bitmap_tree_unlink_element,
6009 bitmap_tree_find_element): New functions for splay-tree bitmap
6011 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
6012 Renamed and moved, see above entries.
6013 (bitmap_tree_listify_from): New function to convert part of a splay
6014 tree bitmap to a linked-list bitmap.
6015 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
6016 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
6017 (bitmap_find_bit): Remove.
6018 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
6019 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
6020 Handle splay tree bitmaps.
6021 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
6022 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
6023 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
6024 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
6025 bitmap_intersect_compl_p, bitmap_ior_and_compl,
6026 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
6027 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
6028 corresponding changes to use linked-list specific bitmap_element
6029 manipulation functions as applicable for efficiency.
6030 (bitmap_tree_to_vec): New function.
6031 (debug_bitmap_elt_file): New function split out from ...
6032 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
6033 (bitmap_print): Likewise.
6035 PR tree-optimization/63155
6036 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
6038 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
6041 2018-10-22 Martin Liska <mliska@suse.cz>
6043 PR tree-optimization/87686
6045 2018-08-29 Martin Liska <mliska@suse.cz>
6047 * tree-switch-conversion.c (switch_conversion::expand):
6048 Strenghten assumption about gswitch statements.
6050 2018-10-22 Martin Liska <mliska@suse.cz>
6052 * ipa-icf.c (sem_item::compare_attributes): Remove.
6053 (sem_item::compare_referenced_symbol_properties): Use
6054 attribute_list_equal instead.
6055 (sem_function::equals_wpa): Likewise.
6056 * ipa-icf.h: Remove compare_attributes.
6058 2018-10-22 Richard Biener <rguenther@suse.de>
6061 * mem-stats.h (mem_usage::operator==): Fix pasto.
6063 2018-10-22 Richard Biener <rguenther@suse.de>
6065 PR tree-optimization/87640
6066 * tree-vrp.c (set_value_range_with_overflow): Decompose
6068 (extract_range_from_binary_expr_1): Adjust.
6070 2018-10-22 Martin Jambor <mjambor@suse.cz>
6072 * tree-eh.h (stmt_could_throw_p): Add function parameter.
6073 (stmt_can_throw_external): Likewise.
6074 (stmt_can_throw_internal): Likewise.
6075 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
6076 (lower_eh_constructs_2): Likewise.
6077 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
6078 (stmt_can_throw_external): Likewise.
6079 (stmt_can_throw_internal): Likewise.
6080 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
6081 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
6082 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
6083 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
6084 (pass_lower_eh_dispatch::execute): Pass cfun to
6085 stmt_can_throw_external.
6086 (cleanup_empty_eh): Likewise.
6087 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
6088 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
6089 stmt_can_throw_external instead of pushing it to cfun.
6090 (symbol_table::create_edge): Likewise.
6091 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
6092 stmt_can_throw_internal.
6093 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
6094 to stmt_could_throw_p.
6095 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
6096 stmt_can_throw_internal.
6097 (pass_store_merging::execute): Likewise.
6098 * gimple-ssa-strength-reduction.c
6099 (find_candidates_dom_walker::before_dom_children): Pass cfun to
6101 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
6102 stmt_can_throw_internal.
6103 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
6104 to stmt_can_throw_external.
6105 (check_stmt): Pass cfun to stmt_could_throw_p.
6106 (check_stmt): Pass cfun to stmt_can_throw_external.
6107 (pass_nothrow::execute): Likewise.
6108 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
6109 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
6110 stmt_can_throw_internal.
6111 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
6112 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
6113 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
6114 * tree-complex.c (expand_complex_libcall): Pass cfun to
6115 stmt_could_throw_p and to stmt_can_throw_internal.
6116 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
6117 * tree-inline.c (copy_edges_for_bb): Likewise.
6118 (maybe_move_debug_stmts_to_successors): Likewise.
6119 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
6121 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
6122 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
6123 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
6124 stmt_can_throw_internal.
6125 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
6126 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
6128 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
6129 stmt_can_throw_internal.
6130 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
6131 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
6133 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
6134 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
6135 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
6136 (convert_mult_to_fma_1): Likewise.
6137 (convert_to_divmod): Likewise.
6138 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
6139 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
6140 * tree-ssa-propagate.c
6141 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
6142 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
6143 (maybe_optimize_range_tests): Likewise.
6144 (linearize_expr_tree): Likewise.
6145 (reassociate_bb): Likewise.
6146 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
6147 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
6148 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
6149 (handle_char_store): Likewise.
6150 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
6151 stmt_can_throw_internal.
6152 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
6154 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
6155 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
6156 (vectorizable_simd_clone_call): Likewise.
6157 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
6158 (gimple_stringop_fixed_value): Likewise.
6160 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
6162 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
6163 literal pool references.
6164 (s390_check_qrst_address): Adapt to the new behavior of
6165 s390_loadrelative_operand_p ().
6167 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
6170 * config/i386/sse.md (*andnot<mode>3_bcst): New.
6172 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
6175 * config/i386/sse.md (*<code><mode>3_bcst): New.
6177 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
6180 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
6181 V4DI, V16SI and V8DI.
6182 (*sub<mode>3<mask_name>_bcst): New.
6183 (*add<mode>3<mask_name>_bcst): Likewise.
6185 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
6186 Jinsong Ji <jji@us.ibm.com>
6188 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
6189 __m64 with __vector unsigned long long for compatibility.
6190 (_mm_movemask_epi8): Likewise.
6191 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
6192 (_mm_cvttps_pi32): Likewise.
6193 (_mm_cvtpi32_ps): Likewise.
6194 (_mm_cvtps_pi16): Likewise.
6195 (_mm_loadh_pi): Likewise.
6196 (_mm_storeh_pi): Likewise.
6197 (_mm_movehl_ps): Likewise.
6198 (_mm_movelh_ps): Likewise.
6199 (_mm_loadl_pi): Likewise.
6200 (_mm_storel_pi): Likewise.
6201 (_mm_movemask_ps): Likewise.
6202 (_mm_shuffle_pi16): Likewise.
6204 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6207 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
6208 __builtin_ia32_vfnmsubpd512_mask.
6209 (_mm512_mask_fnmsub_round_pd): Likewise.
6210 (_mm512_fnmsub_pd): Likewise.
6211 (_mm512_mask_fnmsub_pd): Likewise.
6212 (_mm512_maskz_fnmsub_round_pd): Use
6213 __builtin_ia32_vfnmsubpd512_maskz.
6214 (_mm512_maskz_fnmsub_pd): Likewise.
6215 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
6216 (_mm512_mask_fnmsub_round_ps): Likewise.
6217 (_mm512_fnmsub_ps): Likewise.
6218 (_mm512_mask_fnmsub_ps): Likewise.
6219 (_mm512_maskz_fnmsub_round_ps): Use
6220 __builtin_ia32_vfnmsubps512_maskz.
6221 (_mm512_maskz_fnmsub_ps): Likewise.
6222 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
6223 __builtin_ia32_vfnmsubpd256_mask.
6224 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
6225 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
6226 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
6227 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
6228 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
6229 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
6230 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
6231 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
6232 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
6233 __builtin_ia32_vfnmsubpd.
6234 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
6235 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
6236 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
6237 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
6238 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
6239 * config/i386/i386-builtin.def: Add
6240 __builtin_ia32_vfnmsubpd256_mask,
6241 __builtin_ia32_vfnmsubpd256_maskz,
6242 __builtin_ia32_vfnmsubpd128_mask,
6243 __builtin_ia32_vfnmsubpd128_maskz,
6244 __builtin_ia32_vfnmsubps256_mask,
6245 __builtin_ia32_vfnmsubps256_maskz,
6246 __builtin_ia32_vfnmsubps128_mask,
6247 __builtin_ia32_vfnmsubps128_maskz,
6248 __builtin_ia32_vfnmsubpd512_mask,
6249 __builtin_ia32_vfnmsubpd512_maskz,
6250 __builtin_ia32_vfnmsubps512_mask,
6251 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
6252 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
6253 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
6254 __builtin_ia32_vfnmsubpd256.
6255 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
6256 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
6257 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
6259 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
6261 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
6263 (fmai_vmfnmsub_<mode><round_name>): Likewise.
6265 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6268 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
6269 __builtin_ia32_vfnmaddpd512_mask.
6270 (_mm512_mask_fnmadd_round_pd): Likewise.
6271 (_mm512_fnmadd_pd): Likewise.
6272 (_mm512_mask_fnmadd_pd): Likewise.
6273 (_mm512_maskz_fnmadd_round_pd): Use
6274 __builtin_ia32_vfnmaddpd512_maskz.
6275 (_mm512_maskz_fnmadd_pd): Likewise.
6276 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
6277 (_mm512_mask_fnmadd_round_ps): Likewise.
6278 (_mm512_fnmadd_ps): Likewise.
6279 (_mm512_mask_fnmadd_ps): Likewise.
6280 (_mm512_maskz_fnmadd_round_ps): Use
6281 __builtin_ia32_vfnmaddps512_maskz.
6282 (_mm512_maskz_fnmadd_ps): Likewise.
6283 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
6284 __builtin_ia32_vfnmaddpd256_mask.
6285 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
6286 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
6287 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
6288 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
6289 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
6290 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
6291 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
6292 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
6293 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
6294 __builtin_ia32_vfnmaddpd.
6295 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
6296 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
6297 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
6298 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
6299 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
6300 * config/i386/i386-builtin.def: Add
6301 __builtin_ia32_vfnmaddpd256_mask,
6302 __builtin_ia32_vfnmaddpd256_maskz,
6303 __builtin_ia32_vfnmaddpd128_mask,
6304 __builtin_ia32_vfnmaddpd128_maskz,
6305 __builtin_ia32_vfnmaddps256_mask,
6306 __builtin_ia32_vfnmaddps256_maskz,
6307 __builtin_ia32_vfnmaddps128_mask,
6308 __builtin_ia32_vfnmaddps128_maskz,
6309 __builtin_ia32_vfnmaddpd512_mask,
6310 __builtin_ia32_vfnmaddpd512_maskz,
6311 __builtin_ia32_vfnmaddps512_mask,
6312 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
6313 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
6314 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
6315 __builtin_ia32_vfnmaddpd256.
6316 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
6317 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
6318 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
6320 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
6322 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
6324 (fmai_vmfnmadd_<mode><round_name>): Likewise.
6326 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6329 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
6330 __builtin_ia32_vfmsubpd512_mask.
6331 (_mm512_mask_fmsub_round_pd): Likewise.
6332 (_mm512_fmsub_pd): Likewise.
6333 (_mm512_mask_fmsub_pd): Likewise.
6334 (_mm512_maskz_fmsub_round_pd): Use
6335 __builtin_ia32_vfmsubpd512_maskz.
6336 (_mm512_maskz_fmsub_pd): Likewise.
6337 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
6338 (_mm512_mask_fmsub_round_ps): Likewise.
6339 (_mm512_fmsub_ps): Likewise.
6340 (_mm512_mask_fmsub_ps): Likewise.
6341 (_mm512_maskz_fmsub_round_ps): Use
6342 __builtin_ia32_vfmsubps512_maskz.
6343 (_mm512_maskz_fmsub_ps): Likewise.
6344 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
6345 __builtin_ia32_vfmsubpd256_mask.
6346 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
6347 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
6348 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
6349 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
6350 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
6351 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
6352 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
6353 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
6354 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
6355 __builtin_ia32_vfmsubpd.
6356 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
6357 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
6358 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
6359 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
6360 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
6361 * config/i386/i386-builtin.def: Add
6362 __builtin_ia32_vfmsubpd256_mask,
6363 __builtin_ia32_vfmsubpd256_maskz,
6364 __builtin_ia32_vfmsubpd128_mask,
6365 __builtin_ia32_vfmsubpd128_maskz,
6366 __builtin_ia32_vfmsubps256_mask,
6367 __builtin_ia32_vfmsubps256_maskz,
6368 __builtin_ia32_vfmsubps128_mask,
6369 __builtin_ia32_vfmsubps128_maskz,
6370 __builtin_ia32_vfmsubpd512_mask,
6371 __builtin_ia32_vfmsubpd512_maskz,
6372 __builtin_ia32_vfmsubps512_mask,
6373 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
6374 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
6375 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
6376 __builtin_ia32_vfmsubpd256.
6377 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
6378 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
6379 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
6381 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
6383 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
6385 (fmai_vmfmsub_<mode><round_name>): Likewise.
6387 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6389 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
6390 Remove plus. Renamed to ...
6391 (*sub<mode>3<mask_name>_bcst): This.
6392 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
6393 (*add<mode>3<mask_name>_bcst): This.
6395 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6398 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
6400 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6403 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
6404 (_mm_or_epi32): Likewise.
6405 (_mm256_xor_epi32): Likewise.
6406 (_mm_xor_epi32): Likewise.
6407 (_mm256_or_epi64): Likewise.
6408 (_mm_or_epi64): Likewise.
6409 (_mm256_xor_epi64): Likewise.
6410 (_mm_xor_epi64): Likewise.
6412 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
6415 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
6417 2018-10-20 Jakub Jelinek <jakub@redhat.com>
6420 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
6422 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
6424 * doc/ux.texi: Move @section directly after @node.
6426 2018-10-19 Jakub Jelinek <jakub@redhat.com>
6430 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
6431 depend closely nested inside of loop with ordered clause with
6434 2018-10-19 David Malcolm <dmalcolm@redhat.com>
6436 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
6437 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
6438 * doc/ux.texi: New file.
6440 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
6442 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
6443 be the first CR field allocated.
6445 2018-10-19 Richard Biener <rguenther@suse.de>
6448 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
6449 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
6451 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
6454 * config/i386/sse.md
6455 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
6456 (*add<mode>3<mask_name>_bcst_2): Likewise.
6458 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
6460 * config/i386/sse.md
6461 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
6462 Replace nonimmediate_operand with register_operand.
6463 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
6465 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
6468 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
6470 PR rtl-optimization/87596
6471 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
6472 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
6473 for instructions in FROM..TO range.
6475 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
6477 * cfgexpand.c (expand_one_var): Use specific wording in error message
6478 for non-local frame variables.
6479 * stor-layout.c (layout_decl): Do not issue a warning for them.
6481 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
6483 * haifa-sched.c (priority): Add force_recompute parameter.
6484 (apply_replacement): Call priority () with force_recompute = true.
6485 (restore_pattern): Likewise.
6487 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
6489 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
6490 HOST_BITS_PER_WIDE_INT.
6491 (test_vector_ops_duplicate): Likewise.
6493 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
6496 * config/i386/sse.md (VF_AVX512): New.
6497 (avx512bcst): Likewise.
6498 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
6500 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
6502 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
6505 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
6507 * doc/invoke.texi (-dumpversion): Improve grammar.
6508 (-dumpfullversion): Make more consistent with -dumpversion.
6510 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
6512 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
6513 Set JUMP_LABEL to the jump insn.
6514 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
6515 Predict emitted jump and add label to jump insn.
6517 2018-10-18 David Malcolm <dmalcolm@redhat.com>
6519 PR tree-optimization/87562
6520 * input.c (get_substring_ranges_for_loc): Use
6521 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
6522 getting the linemap for the endpoint. Verify that it's either
6523 in the same linemap as the start point's spelling location, or
6524 at least in the same file.
6526 2018-10-18 Richard Biener <rguenther@suse.de>
6528 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
6529 feed width-specific load/store costs through ix86_vec_cost.
6530 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
6531 (k8_cost): Likewise.
6532 (bdver_cost): Likewise.
6533 (znver1_cost): Likewise.
6534 (btver1_cost): Likewise.
6535 (btver2_cost): Likewise.
6537 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
6539 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
6540 to simplify subreg of vec_merge.
6542 2018-10-18 Richard Biener <rguenther@suse.de>
6544 * config/i386/i386.c: Fix costing of vector FMA.
6546 2018-10-18 Richard Biener <rguenther@suse.de>
6548 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
6550 (ix86_builtin_vectorization_cost): For vec_construct properly
6551 cost insertion into SSE regs.
6552 (...): Adjust calls to ix86_vec_cost.
6554 2018-10-18 Richard Biener <rguenther@suse.de>
6558 2018-02-07 Richard Biener <rguenther@suse.de>
6560 PR tree-optimization/84204
6561 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
6564 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
6567 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
6569 (test_vector_ops_duplicate): Add test for a scalar subreg of a
6570 VEC_MERGE of a VEC_DUPLICATE.
6572 2018-10-17 Joseph Myers <joseph@codesourcery.com>
6574 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
6575 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
6576 * doc/standards.texi (C Language): Document C2X.
6577 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
6578 (rl78_option_override): Handle "GNU C2X" language name.
6580 2018-10-17 Joseph Myers <joseph@codesourcery.com>
6582 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
6583 Document C17 as published in 2018.
6585 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
6588 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
6589 bail out if both sides do not have the same storage order.
6591 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
6593 * bitmap.c (bitmap_head::dump): New.
6594 * bitmap.h (bitmap_head): Add dump().
6595 * gimple-ssa-evrp-analyze.c
6596 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
6597 (evrp_range_analyzer::set_ssa_range_info): Same.
6598 (evrp_range_analyzer::record_ranges_from_phis): Same.
6599 (evrp_range_analyzer::record_ranges_from_stmt): Same.
6600 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
6601 * gimple-ssa-sprintf.c (get_int_range): Same.
6602 (format_integer): Same.
6603 (sprintf_dom_walker::handle_gimple_call): Same.
6604 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
6605 (ipcp_vr_lattice::top_p): Same.
6606 (ipcp_vr_lattice::bottom_p): Same.
6607 (ipcp_vr_lattice::set_to_bottom): Same.
6608 (ipa_vr_operation_and_type_effects): Same.
6609 (propagate_vr_across_jump_function): Same.
6610 (ipcp_store_vr_results): Same.
6611 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
6612 (ipa_print_node_jump_functions_for_edge): Same.
6613 (ipa_get_value_range): Same.
6614 (ipa_compute_jump_functions_for_edge): Same.
6615 (ipa_write_jump_function): Same.
6616 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
6617 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
6619 * vr-values.c (set_value_range_to_nonnegative): Same.
6620 (set_value_range_to_truthvalue): Same.
6621 (vr_values::get_value_range): Same.
6622 (vr_values::set_defs_to_varying): Same.
6623 (vr_values::update_value_range): Same.
6624 (symbolic_range_based_on_p): Same.
6625 (vr_values::op_with_boolean_value_range_p): Same.
6626 (vr_values::extract_range_for_var_from_comparison_expr): Same.
6627 (vr_values::extract_range_from_ssa_name): Same.
6628 (vr_values::extract_range_from_binary_expr): Same.
6629 (vr_values::extract_range_from_unary_expr): Same.
6630 (vr_values::extract_range_from_cond_expr): Same.
6631 (vr_values::extract_range_from_comparison): Same.
6632 (vr_values::check_for_binary_op_overflow): Same.
6633 (vr_values::extract_range_basic): Same.
6634 (vr_values::extract_range_from_assignment): Same.
6635 (compare_ranges): Same.
6636 (compare_range_with_value): Same.
6637 (vr_values::adjust_range_with_scev): Same.
6638 (vrp_valueize): Same.
6639 (vrp_valueize_1): Same.
6640 (vr_values::get_vr_for_comparison): Same.
6641 (vr_values::compare_name_with_value): Same.
6642 (vr_values::compare_names): Same.
6643 (vr_values::vrp_evaluate_conditional): Same.
6644 (find_case_label_ranges): Same.
6645 (vr_values::vrp_visit_switch_stmt): Same.
6646 (vr_values::extract_range_from_phi_node): Same.
6647 (vr_values::simplify_div_or_mod_using_ranges): Same.
6648 (vr_values::simplify_bit_ops_using_ranges): Same.
6649 (test_for_singularity): Same.
6650 (range_fits_type_p): Same.
6651 (vr_values::simplify_cond_using_ranges_1): Same.
6652 (vr_values::simplify_switch_using_ranges): Same.
6653 (vr_values::simplify_float_conversion_using_ranges): Same.
6654 (vr_values::two_valued_val_range_p): Same.
6655 (vr_values::add_equivalence): Move to value_range::equiv_add.
6656 * vr-values.h (vr_values::add_equivalence): Remove.
6657 (VR_INITIALIZER): Remove.
6658 * tree-vrp.c (value_range::set): New.
6659 (value_range::equiv_add): New.
6660 (value_range::value_range): New.
6661 (value_range::deep_copy): New.
6662 (value_range::check): New.
6663 (value_range::equal_p): New.
6664 (value_range::ignore_equivs_equal_p): New.
6665 (value_range::operator==): New.
6666 (value_range::operator!=): New.
6667 (value_range::symbolic_p): New.
6668 (value_range::numeric_p): New.
6669 (value_range::set_undefined): New.
6670 (value_range::set_varying): New.
6671 (value_range::may_contain_p): New.
6672 (value_range::equiv_clear): New.
6673 (value_range::singleton_p): New.
6674 (value_range::intersect): New.
6675 (value_range::dump): New.
6676 (value_range::set_and_canonicalize): New.
6677 (set_value_range): Adjust for value_range API.
6678 (set_value_range_to_undefined): Same.
6679 (set_value_range_to_varying): Same.
6680 (set_and_canonicalize_value_range): Same.
6681 (set_value_range_to_nonnull): Same.
6682 (set_value_range_to_null): Same.
6683 (range_is_null): Same.
6684 (range_is_nonnull): Same.
6685 (range_int_cst_p): Same.
6686 (range_int_cst_singleton_p): Same.
6687 (symbolic_range_p): Same.
6688 (range_includes_zero_p): Same.
6689 (value_range_constant_singleton): Same.
6690 (vrp_set_zero_nonzero_bits): Same.
6691 (ranges_from_anti_range): Same.
6692 (extract_range_into_wide_ints): Same.
6693 (extract_range_from_multiplicative_op): Same.
6694 (set_value_range_with_overflow): Same.
6695 (extract_range_from_binary_expr_1): Same.
6696 (extract_range_from_unary_expr): Same.
6697 (dump_value_range): Same.
6698 (debug_value_range): Same.
6699 (vrp_prop::check_array_ref): Same.
6700 (vrp_prop::check_mem_ref): Same.
6701 (vrp_prop::vrp_initialize): Same.
6702 (vrp_prop::visit_stmt): Same.
6703 (intersect_ranges): Same.
6704 (vrp_prop::visit_phi): Same.
6705 (vrp_prop::vrp_finalize): Same.
6706 (determine_value_range_1): Same.
6707 (determine_value_range): Same.
6708 (vrp_intersect_ranges_1): Rename to...
6709 (vrp_intersect_1): this.
6710 (vrp_intersect_ranges): Rename to...
6711 (value_range::intersect_helper): ...this.
6712 (vrp_meet_1): Rename to...
6713 (value_range::union_helper): ...this.
6714 (vrp_meet): Rename to...
6715 (value_range::union_): ...this.
6716 (copy_value_range): Remove.
6717 * tree-vrp.h (struct value_range): Rewrite into a proper class.
6718 (value_range::vrtype): New.
6719 (value_range::type): New.
6720 (value_range::equiv): New.
6721 (value_range::min): New.
6722 (value_range::max): New.
6723 (value_range::varying_p): New.
6724 (value_range::undefined_p): New.
6725 (value_range::null_p): New.
6726 (value_range::equiv_add): New.
6727 (copy_value_range): Remove.
6729 2018-10-17 David Malcolm <dmalcolm@redhat.com>
6731 * Makefile.in (SELFTEST_TARGETS): New.
6732 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
6733 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
6734 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
6736 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
6737 (selftest-c++-gdb, selftest-c++-valgrind): Move to
6739 * configure: Regenerate.
6740 * configure.ac (selftest_languages): New.
6742 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
6744 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
6745 overflow wraps argument.
6746 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
6747 wide_int_range_multiplicative_op.
6748 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
6749 overflow wraps argument.
6750 (wide_int_range_multiplicative_op): Same.
6751 (wide_int_range_lshift): Same.
6752 (wide_int_range_div): Same.
6753 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
6754 (wide_int_range_lshift): Same.
6755 (wide_int_range_div): Same.
6757 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
6759 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
6760 use sign as argument.
6761 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
6762 wide_int_range_shift_undefined_p.
6764 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6766 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
6768 (@despeculate_copy<ALLI_TI:mode>): ... This.
6769 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
6772 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6774 * config.gcc: Obsolete *-*-solaris2.10*.
6775 * doc/install.texi (Specific, *-*-solaris2*): Document it.
6777 2018-10-12 Jeff Law <law@redhat.com>
6779 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
6780 reg + sym +- const_int addressing modes.
6782 2018-10-15 David Malcolm <dmalcolm@redhat.com>
6784 * common.opt (fdiagnostics-minimum-margin-width=): New option.
6785 * diagnostic-show-locus.c (layout::layout): Apply the minimum
6787 (layout::start_annotation_line): Only print up to 3 of the
6788 margin character, to avoid touching the left-hand side.
6789 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
6790 minimum margin width, as set by test_diagnostic_context's ctor.
6791 (selftest::test_fixit_insert_containing_newline): Likewise.
6792 (selftest::test_fixit_insert_containing_newline_2): Likewise.
6793 (selftest::test_line_numbers_multiline_range): Clear
6794 dc.min_margin_width.
6795 * diagnostic.c (diagnostic_initialize): Initialize
6797 * diagnostic.h (struct diagnostic_context): Add field
6799 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
6800 * opts.c (common_handle_option): Handle
6801 OPT_fdiagnostics_minimum_margin_width_.
6802 * selftest-diagnostic.c
6803 (selftest::test_diagnostic_context::test_diagnostic_context):
6804 Initialize min_margin_width to 6.
6805 * toplev.c (general_init): Initialize global_dc->min_margin_width.
6807 2018-10-15 David Malcolm <dmalcolm@redhat.com>
6809 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
6810 Fix usage of "error_at_rich_loc" in the comment.
6812 2018-10-15 Renlin Li <renlin.li@arm.com>
6815 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
6816 if-conversioned loop when it contains ifn with types not
6817 supported by backend.
6818 * internal-fn.c (expand_direct_optab_fn): Add an assert.
6819 (direct_internal_fn_supported_p): New helper function.
6820 * internal-fn.h (direct_internal_fn_supported_p): Declare.
6822 2018-10-15 Jakub Jelinek <jakub@redhat.com>
6825 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
6826 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
6827 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
6829 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
6831 PR tree-optimization/87022
6832 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
6833 bits in dist vector rather than the first one.
6835 2018-10-15 Richard Biener <rguenther@suse.de>
6838 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
6839 (visit_loadstore): When a used restrict tag escaped verify that
6840 the points-to solution of "other" pointers do not include
6842 (compute_dependence_clique): If a used restrict tag escaped
6843 communicated that down to visit_loadstore.
6845 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
6847 * config/s390/s390.c (s390_expand_vec_init): Force vector element
6848 into reg if it isn't a general operand.
6850 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
6853 * config/i386/sse.md (*vec_dupv2di): Add register source to
6856 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
6859 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
6860 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
6861 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
6862 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
6864 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
6866 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
6867 (notice_args_size): Set it in the current trace if no insn that can
6868 throw internally has been seen yet.
6869 (connect_traces): When connecting args_size between traces, allow the
6870 incoming values not to match if there is an insn setting it before the
6871 first insn that can throw internally; in that case, force the creation
6872 of a CFI note on this latter insn.
6874 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
6876 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
6877 the base class, because using the injected-class-name was not clearly
6878 specified until DR 176.
6880 2018-10-12 Paul Koning <ni1d@arrl.net>
6882 * config/pdp11/pdp11.md (doloop_end): New expander.
6883 (doloop_end_insn): renamed from "doloop_end".
6884 (addqi3): New pattern.
6885 (subqi3): New pattern.
6886 * config/pdp11/predicates.md (incdec_operand): New predicate.
6888 2018-10-12 Yury Gribov <tetra2005@gmail.com>
6891 * real.c (format_helper::can_represent_integral_type_p): New function
6892 * real.h (format_helper::can_represent_integral_type_p): Ditto.
6893 * match.pd: New pattern.
6895 2018-10-12 Alexandre Oliva <oliva@adacore.com>
6897 * configure.ac: Introduce --enable-large-address-aware
6898 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
6899 * doc/install.texi: Document it.
6900 * configure, config.in: Rebuilt.
6901 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
6902 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
6903 (LINK_SPEC): Insert it.
6904 * config/i386/mingw-w64.h: Likewise.
6906 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
6908 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
6910 PR rtl-optimization/87600
6911 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
6913 2018-10-12 Paul Koning <ni1d@arrl.net>
6915 * doc/md.texi (doloop_end): Document that the pattern code may
6916 need to check operand mode.
6918 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
6920 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
6921 to zero-extend between int and floating-point registers.
6922 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
6923 ldp into floating-point registers. Add type and arch attributes.
6924 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
6925 Use f_loads for type attribute.
6927 2018-10-11 Martin Sebor <msebor@redhat.com>
6929 * doc/extend.texi (attribute packed): Correct typos.
6931 2018-10-11 Martin Sebor <msebor@redhat.com>
6933 * doc/extend.texi (attribute flatten): Mention interaction with
6936 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
6939 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
6940 Set new_decl virtual flag to zero.
6942 2018-10-11 Martin Sebor <msebor@redhat.com>
6945 * doc/extend.texi (attribute format_arg): Discuss using multiple
6946 attributes on a single function.
6948 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
6950 PR tree-optimization/86829
6951 * match.pd (sin (atan (x))): New simplification rules.
6952 (cos (atan (x))): Likewise.
6953 * real.c (build_sinatan_real): New function.
6954 * real.h (build_sinatan_real): Prototype.
6956 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
6958 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
6960 (fold_mergeeo_helper): New helper function.
6961 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
6962 intrinsics. Correct some whitespace indentation issues.
6964 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
6967 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
6968 Use HOST_WIDE_INT_1U for shift.
6970 2018-10-11 Doug Rupp <rupp@adacore.com>
6971 Olivier Hainque <hainque@adacore.com>
6973 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
6974 Pass --relax to the linker for RTPs.
6975 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
6977 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
6978 Jan Hubicka <jh@suse.cz>
6979 Martin Jambor <mjambor@suse.cz>
6981 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
6982 the same elements are repeated rather than printing all of them.
6983 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
6984 "repeated" elements.
6985 * read-rtl-function.c (test_loading_repeat): New function.
6986 (read_rtl_function_c_tests): Call test_loading_repeat.
6987 * rtl-tests.c (test_dumping_repeat): New function.
6988 (rtl_tests_c_tests): Call test_dumping_repeat.
6990 2018-10-11 Richard Biener <rguenther@suse.de>
6992 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
6993 bdver?_cost): Unify to ...
6994 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
6995 * config/i386/i386.c (processor_cost_table): Adjust.
6997 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
7000 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
7001 the thunk when expanding to GIMPLE.
7003 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
7006 * varasm.c (mergeable_string_section): Don't try to move zero-length
7007 strings to the merge section.
7009 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
7012 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
7014 2018-10-10 Jakub Jelinek <jakub@redhat.com>
7017 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
7018 to special_args set.
7020 2018-10-10 Richard Biener <rguenther@suse.de>
7022 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
7023 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
7024 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
7025 and recursing and pattern terminating the recursion on SSE
7026 vector width using ix86_expand_reduc.
7027 (reduc_sminmax_scal_<mode>): Split into part reducing to half
7028 width and recursing and SSE2 vector variant doing the final
7029 reduction with ix86_expand_reduc.
7030 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
7031 with terminating the recursion at AVX level, splitting that
7034 2018-10-09 David Malcolm <dmalcolm@redhat.com>
7036 * genmatch.c (error_cb): Rename to...
7037 (diagnostic_cb): ...this, converting int params to enums.
7038 (fatal_at): Update for renaming.
7039 (warning_at): Likewise.
7041 * input.c (selftest::ebcdic_execution_charset::apply):
7042 Update for renaming of...
7043 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
7045 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
7046 converting level and reason to enums.
7047 (class selftest::lexer_error_sink): Rename to...
7048 (class selftest::lexer_test_options): ...this, renaming field
7049 "m_errors" to "m_diagnostics".
7050 (selftest::lexer_test_options::apply): Update for renaming of...
7051 (selftest::lexer_test_options::on_error): ...this, renaming to...
7052 (selftest::lexer_test_options::on_diagnostic): ...this
7053 converting level and reason to enums.
7054 (selftest::test_lexer_string_locations_raw_string_unterminated):
7055 Update for renamings.
7056 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
7059 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
7061 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
7062 * config/rs6000/pmmintrin.h: New file.
7064 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
7066 PR tree-optimization/86659
7067 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
7069 2018-10-09 Richard Biener <rguenther@suse.de>
7071 PR tree-optimization/63155
7072 * tree-ssa-structalias.c: Include tree-ssa.h.
7073 (get_constraint_for_ssa_var): For undefs return nothing_id.
7074 (find_func_aliases): Cleanup PHI handling.
7076 2018-10-09 Richard Biener <rguenther@suse.de>
7078 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
7081 2018-10-09 Martin Liska <mliska@suse.cz>
7083 * asan.c (asan_emit_stack_protection): If a stack variable
7084 is located in a same file as current function, then emit
7085 line info into variable definition string.
7087 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
7089 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
7092 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
7094 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
7097 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
7099 PR tree-optimization/86659
7100 * gimple-match.h (struct gimple_match_op): Add reverse field.
7101 (gimple_match_op::set_op): New overloaded method.
7102 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
7103 the REF_REVERSE_STORAGE_ORDER flag on the value.
7104 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
7105 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
7107 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
7110 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
7111 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
7113 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
7116 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
7117 Defined with __builtin_ia32_vfmaddsubpd512_mask.
7119 2018-10-08 Richard Biener <rguenther@suse.de>
7121 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
7122 cost the same as AVX128 ones.
7124 2018-10-08 Paul Koning <ni1d@arrl.net>
7126 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
7127 (expand_block_move): New function.
7128 * config/pdp11/pdp11.c (output_block_move): Remove.
7129 (expand_block_move): New function.
7130 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
7131 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
7132 (*movmemhi1): Remove.
7134 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
7136 * config/s390/2827.md: Increase latencies for some FP instructions.
7138 2018-10-08 Richard Biener <rguenther@suse.de>
7140 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
7142 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
7143 * tree-vectorizer.h (dump_stmt_cost): Adjust.
7144 (add_stmt_cost): Dump return value of the hook.
7146 2018-10-08 Richard Biener <rguenther@suse.de>
7148 PR tree-optimization/63155
7149 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
7150 (ssa_propagation_engine::ssa_propagate): Remove redundant
7151 bitmap bit clearing.
7153 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
7155 PR rtl-optimization/86939
7156 PR rtl-optimization/87479
7157 * ira.h (non_conflicting_reg_copy_p): New prototype.
7158 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
7159 (make_hard_regno_dead): Don't add conflicts for register
7160 ignore_reg_for_conflicts.
7161 (make_object_dead): Likewise.
7162 (non_conflicting_reg_copy_p): New function.
7163 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
7164 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
7165 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
7166 (make_hard_regno_dead): Don't add conflicts for register
7167 ignore_reg_for_conflicts. Remove special conflict handling of
7168 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
7169 check_pic_pseudo_p and update callers.
7170 (mark_pseudo_dead): Don't add conflicts for register
7171 ignore_reg_for_conflicts.
7172 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
7174 2018-10-05 Andrew Waterman <andrew@sifive.com>
7175 Jim Wilson <jimw@sifive.com>
7177 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
7178 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
7179 new pattern using HONOR_SNANS that emits one extra instruction.
7181 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
7183 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
7184 patterns): Merge SI and DI patterns to a GPR pattern.
7185 (unnamed define_insn and define_split for record form of that): Merge
7186 to a single define_insn_and_split pattern.
7188 2018-10-05 David Malcolm <dmalcolm@redhat.com>
7191 * input.c (expand_location_to_spelling_point): Add param "aspect"
7192 and use rather than hardcoding LOCATION_ASPECT_CARET.
7193 (get_substring_ranges_for_loc): Handle the case of a single token
7194 within a macro expansion.
7195 * input.h (expand_location_to_spelling_point): Add "aspect" param,
7196 defaulting to LOCATION_ASPECT_CARET.
7198 2018-10-05 Paul Koning <ni1d@arrl.net>
7200 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
7201 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
7202 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
7203 (pdp11_guard_type): New function.
7205 2018-10-05 Paul Koning <ni1d@arrl.net>
7207 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
7208 * config/pdp11/pdp11.opt (mfloat32): Remove.
7210 * doc/invoke.texi (pdp11 -mfloat32): Remove:
7211 (pdp11 -mfloat64): Remove.
7213 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
7215 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
7216 (*cmp<mode>_cc_i387): Ditto.
7217 (*cmpu<mode>_cc_i387): Ditto.
7218 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
7219 * config/i386/i386.c (ix86_expand_fp_compare): Remove
7221 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
7222 Emit x86_sahf_1 pattern.
7223 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
7224 (ix86_expand_carry_flag_compare): Ditto.
7226 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
7228 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
7229 to reg_or_0_operand. Add "C" constraint.
7230 (*cmpxf_cc_i387): Ditto.
7231 (*cmp<mode>_i387): Change operand 2 predicate
7232 to nonimm_or_0_operand. Add "C" constraint.
7233 (*cmp<mode>_cc_i387): Ditto.
7234 (*cmp<mode>_0_i387): Remove insn pattern.
7235 (*cmp<mode>_0_cc_i387): Ditto.
7237 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
7239 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
7240 * config/i386/predicates.md (nonimm_or_0_operand): Rename
7241 from vector_move_operand. Update all uses.
7243 2018-10-05 Martin Sebor <msebor@redhat.com>
7245 PR tree-optimization/87490
7246 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
7249 2018-10-05 Richard Biener <rguenther@suse.de>
7251 PR tree-optimization/63155
7252 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
7253 vertical space in dumpfiles.
7254 * tree-ssa-propagate.h
7255 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
7256 * tree-ssa-propagate.c (cfg_blocks_back): New global.
7257 (ssa_edge_worklist_back): Likewise.
7258 (curr_order): Likewise.
7259 (cfg_blocks_get): Remove abstraction.
7260 (cfg_blocks_add): Likewise.
7261 (cfg_blocks_empty_p): Likewise.
7262 (add_ssa_edge): Add to current or next worklist based on
7264 (add_control_edge): Likewise.
7265 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
7267 (ssa_propagation_engine::ssa_propagate): ... here. Unify
7268 iteration from CFG and SSA edge worklist so we process
7269 everything in RPO order, prioritizing forward progress
7271 (ssa_prop_init): Allocate new worklists, do not dump
7273 (ssa_prop_fini): Free new worklists.
7275 2018-10-05 Richard Biener <rguenther@suse.de>
7277 * tree-core.h (tree_block::abstract_flag): Remove.
7278 (tree_block::block_num): Make full 32bits.
7279 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
7280 * tree.h (BLOCK_ABSTRACT): Remove.
7281 * dwarf2out.c (gen_lexical_block_die): Remove dead code
7282 resulting from BLOCK_ABSTRACT being always false.
7283 (gen_inlined_subroutine_die): Likewise.
7284 (gen_block_die): Likewise.
7285 * tree.c (block_ultimate_origin): Likewise.
7286 * tree-pretty-print.c (dump_block_node): Remove code dealing
7287 with BLOCK_ABSTRACT.
7288 * tree-ssa-live.c (dump_scope_block): Likewise.
7289 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
7290 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
7292 2018-10-05 Richard Biener <rguenther@suse.de>
7294 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
7295 is asked for initialize mode to the component mode of the
7298 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
7301 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
7302 assembler for -mavx.
7303 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
7305 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
7308 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
7310 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
7311 to be DImode. When using mffscrn, force the operand to a register.
7313 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
7315 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
7316 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
7317 X87MODEF mode iterator.
7318 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
7319 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
7320 X87MODEF mode iterator.
7322 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
7324 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
7325 -Wno-prio-ctor-dtor.
7327 2018-10-04 David Malcolm <dmalcolm@redhat.com>
7329 * Makefile.in (OBJS): Add opt-problem.o.
7330 * dump-context.h: Include "selftest.h.
7331 (selftest::temp_dump_context): New forward decl.
7332 (class dump_context): Make friend of class
7333 selftest::temp_dump_context.
7334 (dump_context::dump_loc_immediate): New decl.
7335 (class dump_pretty_printer): Move here from dumpfile.c.
7336 (class temp_dump_context): Move to namespace selftest.
7337 (temp_dump_context::temp_dump_context): Add param
7338 "forcibly_enable_dumping".
7339 (selftest::verify_dumped_text):
7340 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
7341 (selftest::verify_item):
7342 (ASSERT_IS_TEXT): Move here from dumpfile.c.
7343 (ASSERT_IS_TREE): Likewise.
7344 (ASSERT_IS_GIMPLE): Likewise.
7345 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
7347 (dump_context::dump_loc_immediate): ...this new function.
7348 (class dump_pretty_printer): Move to dump-context.h.
7349 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
7350 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
7351 (temp_dump_context::temp_dump_context): Move to "selftest"
7352 namespace. Add param "forcibly_enable_dumping", and use it to
7353 conditionalize the use of m_pp;
7354 (selftest::verify_dumped_text): Make non-static.
7355 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
7356 (selftest::verify_item): Make non-static.
7357 (ASSERT_IS_TEXT): Move to dump-context.h.
7358 (ASSERT_IS_TREE): Likewise.
7359 (ASSERT_IS_GIMPLE): Likewise.
7360 (selftest::test_capture_of_dump_calls): Pass "true" for new
7361 param of temp_dump_context.
7362 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
7363 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
7365 * opt-problem.cc: New file.
7366 * opt-problem.h: New file.
7367 * optinfo-emit-json.cc
7368 (selftest::test_building_json_from_dump_calls): Pass "true" for
7369 new param of temp_dump_context.
7370 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
7371 (optinfo::emit_for_opt_problem): New function.
7372 (optinfo::emit): Clarity which emit_item is used.
7373 * optinfo.h (optinfo::get_dump_location): New accessor.
7374 (optinfo::emit_for_opt_problem): New decl.
7375 (optinfo::emit): Make const.
7376 * selftest-run-tests.c (selftest::run_tests): Call
7377 selftest::opt_problem_cc_tests.
7378 * selftest.h (selftest::opt_problem_cc_tests): New decl.
7379 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
7380 bool to opt_result, converting fprintf messages to
7381 opt_result::failure_at calls. Add "stmt" param for use by the
7383 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
7384 (runtime_alias_check_p): Convert return type from bool to
7385 opt_result, converting dump_printf calls to
7386 opt_result::failure_at, using the statement DDR_A for their
7388 (find_data_references_in_stmt): Convert return type from bool to
7389 opt_result, converting "return false" to opt_result::failure_at
7391 * tree-data-ref.h: Include "opt-problem.h".
7392 (dr_analyze_innermost): Convert return type from bool to opt_result,
7393 and add a const gimple * param.
7394 (find_data_references_in_stmt): Convert return type from bool to
7396 (runtime_alias_check_p): Likewise.
7397 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
7398 dr_analyze_innermost.
7399 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
7400 Convert return type from bool to opt_result, adding a message for
7401 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
7402 (vect_analyze_data_ref_dependence): Convert return type from bool
7403 to opt_result. Change sense of return type from "false"
7404 effectively meaning "no problems" to "false" meaning a problem,
7405 so that "return false" becomes "return opt_result::success".
7406 Convert "return true" calls to opt_result::failure_at, using
7407 the location of statement A rather than vect_location.
7408 (vect_analyze_data_ref_dependences): Convert return type from bool
7410 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
7411 calls to opt_result::failure_at, using the stmt location rather
7413 (vect_verify_datarefs_alignment): Convert return type from bool
7415 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
7416 into multiple more-tightly-scoped copies.
7417 (vect_analyze_data_refs_alignment): Convert return type from bool
7419 (vect_analyze_data_ref_accesses): Likewise, converting a
7420 "return false" to a "return opt_result::failure_at", adding a
7422 (vect_prune_runtime_alias_test_list): Convert return type from
7423 bool to opt_result, converting dump_printf_loc to
7424 opt_result::failure_at. Add a %G to show the pertinent statement,
7425 and use the stmt's location rather than vect_location.
7426 (vect_find_stmt_data_reference): Convert return type from
7427 bool to opt_result, converting dump_printf_loc to
7428 opt_result::failure_at, using stmt's location.
7429 (vect_analyze_data_refs): Convert return type from bool to
7430 opt_result. Convert "return false" to "return
7431 opt_result::failure_at", adding messages as needed.
7432 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
7433 type from bool to opt_result.
7434 (vect_determine_vf_for_stmt): Likewise.
7435 (vect_determine_vectorization_factor): Likewise, converting
7436 dump_printf_loc to opt_result::failure_at, using location of phi
7437 rather than vect_location.
7438 (vect_analyze_loop_form_1): Convert return type from bool to
7439 opt_result, converting dump_printf_loc calls, retaining the use of
7441 (vect_analyze_loop_form): Convert return type from loop_vec_info
7442 to opt_loop_vec_info.
7443 (vect_analyze_loop_operations): Convert return type from bool to
7444 opt_result, converting dump_printf_loc calls, using the location
7445 of phi/stmt rather than vect_location where available. Convert
7446 various "return false" to "return opt_result::failure_at" with
7447 "unsupported phi" messages.
7448 (vect_get_datarefs_in_loop): Convert return type from bool to
7449 opt_result. Add a message for the
7450 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
7451 (vect_analyze_loop_2): Convert return type from bool to
7452 opt_result. Ensure "ok" is set to a opt_result::failure_at before
7453 each "goto again;", adding new messages where needed.
7454 Add "unsupported grouped {store|load}" messages.
7455 (vect_analyze_loop): Convert return type from loop_vec_info to
7457 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
7459 * tree-vect-stmts.c (process_use): Likewise, converting
7460 dump_printf_loc call and using stmt location, rather than
7462 (vect_mark_stmts_to_be_vectorized): Likeise.
7463 (vect_analyze_stmt): Likewise, adding a %G.
7464 (vect_get_vector_types_for_stmt): Convert return type from bool to
7465 opt_result, converting dump_printf_loc calls and using stmt
7466 location, rather than vect_location.
7467 (vect_get_mask_type_for_stmt): Convert return type from tree to
7468 opt_tree, converting dump_printf_loc calls and using stmt location.
7469 * tree-vectorizer.c: Include "opt-problem.h.
7470 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
7471 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
7472 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
7473 enabled, use it to report at the top level "couldn't vectorize
7474 loop" followed by the problem.
7475 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
7476 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
7478 (vect_analyze_stmt): Likewise.
7479 (vect_get_vector_types_for_stmt): Likewise.
7480 (tree vect_get_mask_type_for_stmt): Likewise.
7481 (vect_analyze_data_ref_dependences): Likewise.
7482 (vect_enhance_data_refs_alignment): Likewise.
7483 (vect_analyze_data_refs_alignment): Likewise.
7484 (vect_verify_datarefs_alignment): Likewise.
7485 (vect_analyze_data_ref_accesses): Likewise.
7486 (vect_prune_runtime_alias_test_list): Likewise.
7487 (vect_find_stmt_data_reference): Likewise.
7488 (vect_analyze_data_refs): Likewise.
7489 (vect_analyze_loop): Convert return type from loop_vec_info to
7491 (vect_analyze_loop_form): Likewise.
7492 (vect_analyze_slp): Convert return type from bool to opt_result.
7494 2018-10-04 David Malcolm <dmalcolm@redhat.com>
7496 * doc/invoke.texi (-fopt-info): Document new "internals"
7498 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
7499 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
7501 (optinfo_verbosity_options): Add "internals".
7502 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
7503 (dump_context::apply_dump_filter_p): New member function.
7504 (dump_context::dump_loc): Use apply_dump_filter_p rather than
7505 explicitly masking the dump_kind.
7506 (dump_context::begin_scope): Increment the scope depth first. Use
7507 apply_dump_filter_p rather than explicitly masking the dump_kind.
7508 (dump_context::emit_item): Use apply_dump_filter_p rather than
7509 explicitly masking the dump_kind.
7510 (dump_dec): Likewise.
7511 (dump_hex): Likewise.
7512 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
7513 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
7514 (opt_info_switch_p): Update handling of default
7515 MSG_OPTIMIZED_LOCATIONS to cope with default of
7516 MSG_PRIORITY_USER_FACING.
7517 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
7518 masking the dump_kind.
7519 (selftest::test_capture_of_dump_calls): Update test_dump_context
7520 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
7521 than MSG_ALL. Generalize scope test to be run at all four
7522 combinations of with/without MSG_PRIORITY_USER_FACING and
7523 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
7524 for each of the two values.
7525 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
7526 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
7527 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
7528 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
7529 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
7530 with MSG_PRIORITY_*.
7531 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
7532 dump messages as MSG_PRIORITY_USER_FACING.
7533 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
7534 about the interaction with MSG_PRIORITY_*.
7536 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
7538 * varasm.c (output_constant): Add new parameter merge_strings.
7539 Make strings properly zero terminated in merge string sections.
7540 (mergeable_string_section): Don't fail if the last char is non-zero.
7541 (assemble_variable_contents): Handle merge string sections.
7542 (assemble_variable): Likewise.
7543 (assemble_constant_contents): Likewise.
7544 (output_constant_def_contents): Likewise.
7545 (output_constructor_array_range,
7546 output_constructor_regular_field): Adjust call to output_constant.
7547 (output_object_block): Adjust call to assemble_constant_contents
7548 and assemble_variable_contents.
7550 2018-10-04 Martin Liska <mliska@suse.cz>
7553 * cgraphunit.c (process_function_and_variable_attributes):
7554 Warn about a function with alias attribute and a body.
7556 2018-10-04 Martin Liska <mliska@suse.cz>
7559 * multiple_target.c (redirect_to_specific_clone): New function.
7560 (ipa_target_clone): Use it.
7561 * tree-inline.c: Fix comment.
7563 2018-10-04 David Malcolm <dmalcolm@redhat.com>
7565 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
7567 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
7568 (gcc::dump_manager::register_pass): New member function, adapted
7569 from loop body in gcc::pass_manager::register_pass, adding a
7570 call to update_dfi_for_opt_info.
7571 (gcc::dump_manager::opt_info_enable_passes): Store the
7572 -fopt-info options into the new fields. Move the loop
7574 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
7576 * dumpfile.h (struct opt_pass): New forward decl.
7577 (gcc::dump_manager::register_pass): New decl.
7578 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
7579 (class gcc::dump_manager): Add fields "m_optgroup_flags",
7580 "m_optinfo_flags", and "m_optinfo_filename".
7581 * passes.c (gcc::pass_manager::register_pass): Move all of the
7582 dump-handling code to gcc::dump_manager::register_pass.
7584 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
7586 PR rtl-optimization/87466
7587 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
7588 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
7589 * doc/tm.texi: Regenerate.
7590 * ira-lives.c (process_bb_node_lives): Use the new target hook.
7591 * lra-lives.c (process_bb_lives): Likewise.
7592 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
7595 2018-10-04 Tamar Christina <tamar.christina@arm.com>
7597 * params.c (add_params): Fix initialization.
7599 2018-10-04 Martin Liska <mliska@suse.cz>
7601 PR gcov-profile/84107
7602 * tree-profile.c (init_ic_make_global_vars):
7603 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
7604 Come up with new ic_tuple* variables. Emit
7605 __gcov_indirect_call{,_topn} variables.
7606 (gimple_gen_ic_profiler): Access the variable
7608 (gimple_gen_ic_func_profiler): Access
7609 __gcov_indirect_call.callee field.
7610 (gimple_init_gcov_profiler): Use ptr_type_node.
7611 * value-prof.c (gimple_ic): Use ptr_type_node.
7613 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7615 PR tree-optimization/85787
7616 * ipa-pure-const.c (malloc_candidate_p_1): Move most of
7617 malloc_candidate_p into this function and add support for
7618 detecting multiple phis.
7619 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
7621 2018-10-04 Martin Liska <mliska@suse.cz>
7624 * ipa-inline.c (inline_to_all_callers_1):
7625 Call ultimate_alias_target for node being inlined.
7627 2018-10-03 Jeff Law <law@redhat.com>
7629 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
7631 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
7632 * tree.h (get_typenode_from_name): Prototype.
7634 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
7636 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
7637 Change operand 2 predicate to nonimmediate_operand.
7638 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
7640 2018-10-03 Martin Sebor <msebor@redhat.com>
7641 Jeff Law <law@redhat.com>
7643 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
7645 (get_string_length): Detect unterminated arrays.
7646 (format_string): Same.
7647 (format_directive): Warn about unterminated arrays.
7648 (handle_gimple_call): Mark statements with no_warning as needed.
7650 2018-10-03 Jim Wilson <jimw@sifive.com>
7652 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
7653 also define __riscv_abi_rve. Delete trailing white space.
7655 2018-10-03 Paul Koning <ni1d@arrl.net>
7657 Enable LRA register allocator for PDP11.
7658 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
7661 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
7662 * config/pdp11/pdp11.opt (-mlra): New option.
7663 * doc/invoke.texi (PDP-11 Options): Document -mlra.
7665 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
7667 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
7668 (*<absneg:code>extend<mode>xf2): Ditto.
7670 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
7672 PR tree-optimization/87415
7673 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
7676 2018-10-02 Jeff Law <law@redhat.com>
7678 * gimple-fold.c (get_range_strlen): Only set *nonstr when
7679 an unterminated string is discovered. Bubble up range
7680 even for unterminated strings.
7681 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
7682 indicates the string was not terminated via NONSTR.
7684 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
7686 * tree-vrp.c (extract_range_from_unary_expr): Special case all
7687 pointer conversions.
7688 Do not do anything special for anti-ranges.
7690 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
7692 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
7693 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
7695 2018-10-03 Martin Liska <mliska@suse.cz>
7697 PR gcov-profile/86109
7698 * coverage.c (coverage_begin_function): Do not
7699 mark lambdas as artificial.
7700 * tree-core.h (struct GTY): Remove tm_clone_flag
7701 and introduce new lambda_function.
7702 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
7704 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
7707 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
7708 P8_VECTOR and VSX are enabled.
7710 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
7712 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
7713 0x3907 as CPU model number.
7715 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
7717 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
7718 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
7719 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
7720 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
7721 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
7722 * config/s390/s390.md: Likewise. Rename also the cpu attribute
7723 value from arch12 to z14.
7725 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
7727 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
7729 (isinf<mode>2): Ditto.
7731 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
7733 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
7734 before emitting fxam. Perform calculations in XFmode.
7736 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
7738 * match.pd (((X /[ex] A) +- B) * A): New transformation.
7740 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
7743 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
7744 * tree.c (signed_or_unsigned_type_for): Handle complex.
7746 2018-10-02 Jeff Law <law@redhat.com>
7748 * gimple-fold.c (get_range_strlen): Remove dead code.
7750 2018-10-02 Martin Sebor <msebor@redhat.com>
7751 Jeff Law <law@redhat.com>
7753 * builtins.c (unterminated_array): Add new arguments.
7754 If argument is not terminated, bubble up size and exact
7756 (expand_builtin_strnlen): Detect, avoid expanding
7757 and diagnose unterminated arrays.
7758 (c_strlen): Fill in offset of start of unterminated strings.
7759 * builtins.h (unterminated_array): Update prototype.
7761 2018-10-02 Richard Biener <rguenther@suse.de>
7763 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
7764 of haddv4df, first reduce to SSE width and exploit the fact
7765 that we only need element zero with the reduction result.
7766 (reduc_plus_scal_v2df): Likewise.
7768 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
7770 * dojump.h (do_jump): Delete.
7771 (do_jump_1): Likewise.
7772 (split_comparison): Move around.
7773 * dojump.c (do_jump): Make static.
7774 (do_jump_1): Likewise.
7775 (jumpifnot): Move around.
7776 (jumpifnot_1): Likewise.
7778 (jumpif_1): Likewise.
7779 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
7781 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
7783 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
7784 insns in the delay slot and add_insn_after for the jump insn.
7786 2018-10-02 Richard Biener <rguenther@suse.de>
7788 * tree-inline.c (expand_call_inline): Use the location of
7789 the callee declaration for the inline-entry marker.
7790 * final.c (notice_source_line): Remove special-casing of
7791 NOTE_INSN_INLINE_ENTRY.
7793 2018-10-01 Carl Love <cel@us.ibm.com>
7796 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
7797 (__builtin_mtfsb0): New.
7798 (__builtin_mtfsb1): New.
7799 ( __builtin_set_fpscr_rn): New.
7800 (__builtin_set_fpscr_drn): New.
7801 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
7802 (rs6000_expand_set_fpscr_rn_builtin): Add.
7803 (rs6000_expand_set_fpscr_drn_builtin): Add.
7804 (rs6000_expand_builtin): Add case statement entries for
7805 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
7806 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
7807 RS6000_BUILTIN_MFFSL.
7808 (rs6000_init_builtins): Add ftype initialization and def_builtin
7809 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
7810 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
7811 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
7812 rs6000_mffscdrn): Add define_insn.
7813 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
7814 * doc/extend.texi: Add documentation for the builtins.
7816 2018-10-01 Richard Biener <rguenther@suse.de>
7818 PR tree-optimization/87465
7819 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
7820 causing branch miscounts.
7822 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7824 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
7825 aarch64_option_default_param): New.
7826 (params.h): Include.
7827 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
7828 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
7829 stack-clash protection validation code.
7831 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7833 * params.c (validate_param): New.
7834 (add_params): Use it.
7835 (set_param_value): Refactor param validation into validate_param.
7836 (diagnostic.h): Include.
7837 * diagnostic.h (diagnostic_ready_p): New.
7839 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7841 * params.c (set_param_value):
7842 Add index of parameter being validated.
7843 * common/common-target.def (option_validate_param): New.
7844 * common/common-targhooks.h (default_option_validate_param): New.
7845 * common/common-targhooks.c (default_option_validate_param): New.
7846 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
7847 * doc/tm.texi: Regenerate.
7849 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7852 * config/aarch64/aarch64.c (aarch64_override_options_internal):
7853 Add validation for stack-clash parameters and set defaults.
7855 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7858 * configure.ac: Add stack-clash-protection-guard-size.
7859 * doc/install.texi: Document it.
7860 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
7861 * params.def: Update comment for guard-size.
7862 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
7863 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
7864 * configure: Regenerate.
7866 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7869 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
7870 STACK_DYNAMIC_OFFSET): New.
7871 * config/aarch64/aarch64.c (aarch64_layout_frame):
7872 Update outgoing args size.
7873 (aarch64_stack_clash_protection_alloca_probe_range,
7874 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
7876 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7879 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
7881 * target.def (stack_clash_protection_alloca_probe_range): New.
7882 (stack_clash_protection_final_dynamic_probe): Remove.
7883 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
7884 (default_stack_clash_protection_final_dynamic_probe): Remove.
7885 * targhooks.c: Likewise.
7886 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
7887 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
7888 * doc/tm.texi: Regenerate.
7890 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7893 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
7894 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
7895 aarch64_clamp_to_uimm12_shift): New.
7896 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
7897 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
7899 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7902 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
7904 2018-10-01 Jeff Law <law@redhat.com>
7905 Richard Sandiford <richard.sandiford@linaro.org>
7906 Tamar Christina <tamar.christina@arm.com>
7909 * config/aarch64/aarch64.md
7910 (probe_stack_range): Add k (SP) constraint.
7911 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
7912 STACK_CLASH_MAX_UNROLL_PAGES): New.
7913 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
7914 stack probes for stack clash.
7915 (aarch64_allocate_and_probe_stack_space): New.
7916 (aarch64_expand_prologue): Use it.
7917 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
7918 (aarch64_sub_sp): Add emit_move_imm optional param.
7920 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
7922 PR tree-optimization/87261
7923 * match.pd: Remove trailing whitespace.
7924 Add (x & y) | ~(x | y) -> ~(x ^ y),
7925 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
7927 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
7929 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
7931 (ashlsi3_insn): Update instruction constraints.
7932 (ashrsi3_insn): Likewise.
7933 (rotrsi3): Likewise.
7934 (add_shift): Likewise.
7935 * config/arc/constraints.md (Csz): New 32 bit constraint. It
7936 avoids placing in the limm field small constants which, otherwise,
7937 could end into a small instruction.
7939 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
7941 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
7942 destination register is not odd-even.
7943 (umaddsidi4_split): Likewise.
7945 2018-10-01 Richard Biener <rguenther@suse.de>
7947 * tree-inline.c (expand_call_inline): Store origin of fn
7948 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
7949 * tree.c (block_ultimate_origin): Simplify and do some
7952 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
7954 * config/i386/mmx.md (EMMS): New int iterator.
7955 (emms): New int attribute.
7956 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
7957 EMMS int iterator. Explicitly declare clobbers.
7958 (mmx_emms): Remove expander.
7960 * config/i386/predicates.md (emms_operation): Remove predicate.
7961 (vzeroall_pattern): New predicate.
7962 (vzeroupper_pattern): Rename from vzeroupper_operation.
7963 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
7964 vzeroupper_pattern and vzeroall_pattern predicates.
7966 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
7968 PR rtl-optimization/86939
7969 * ira-lives.c (make_hard_regno_born): Rename from this...
7970 (make_hard_regno_live): ... to this. Remove update to conflict
7971 information. Update function comment.
7972 (make_hard_regno_dead): Add conflict information update. Update
7974 (make_object_born): Rename from this...
7975 (make_object_live): ... to this. Remove update to conflict information.
7976 Update function comment.
7977 (make_object_dead): Add conflict information update. Update function
7979 (mark_pseudo_regno_live): Call make_object_live.
7980 (mark_pseudo_regno_subword_live): Likewise.
7981 (mark_hard_reg_dead): Update function comment.
7982 (mark_hard_reg_live): Call make_hard_regno_live.
7983 (process_bb_node_lives): Likewise.
7984 * lra-lives.c (make_hard_regno_born): Rename from this...
7985 (make_hard_regno_live): ... to this. Remove update to conflict
7986 information. Remove now uneeded check_pic_pseudo_p argument.
7987 Update function comment.
7988 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
7989 to conflict information. Update function comment.
7990 (mark_pseudo_live): Remove update to conflict information. Update
7992 (mark_pseudo_dead): Add conflict information update.
7993 (mark_regno_live): Call make_hard_regno_live.
7994 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
7995 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
7997 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
8000 * config/i386/i386.c (construct_container): Use TImode for
8001 BLKmode values in 2 integer registers.
8003 2018-09-29 Jeff Law <law@redhat.com>
8005 * builtins.c (unterminated_array): Pass in c_strlen_data * to
8006 c_strlen rather than just a tree *.
8007 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
8008 Update recursive calls appropriately. If caller did not provide a
8009 suitable data pointer, create a local one. When a non-terminated
8010 string is discovered, bubble up information about the string via the
8011 c_strlen_data object.
8012 * builtins.h (c_strlen): Update prototype.
8013 (c_strlen_data): New structure.
8014 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
8015 For a type 2 call, if c_strlen indicates a non-terminated string
8016 use the length of the non-terminated string.
8017 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
8019 2018-09-29 Jakub Jelinek <jakub@redhat.com>
8022 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
8023 __m512d type for __A argument rather than __m512.
8025 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
8027 * match.pd (simple_comparison): Don't optimize if either operand is
8028 a function pointer when target needs function pointer canonicalization.
8030 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
8032 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
8033 power5 .. power9 to remove indirection.
8034 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
8035 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
8036 ASM_CPU_476_SPEC): Delete.
8037 (ASM_CPU_SPEC): Adjust.
8038 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
8039 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
8041 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
8043 * config.in: Delete HAVE_AS_DCI.
8044 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
8045 * config/rs6000/rs6000.h: Ditto.
8046 * configure.ac: Delete HAVE_AS_DCI.
8047 * configure: Regenerate.
8049 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
8051 * config.in (HAVE_AS_LWSYNC): Delete.
8052 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
8053 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
8055 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
8056 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
8058 * configure.ac: Delete HAVE_AS_LWSYNC.
8059 * configure: Regenerate.
8061 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
8062 Pierre-Marie de Rodat <derodat@adacore.com>
8064 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
8065 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
8066 (cgraph_node::create_thunk): Add indirect_offset parameter.
8067 (thunk_adjust): Likewise.
8068 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
8069 and initialize the corresponding field with it.
8070 (cgraph_node::dump): Dump indirect_offset field.
8071 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
8072 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
8073 (thunk_adjust): Add indirect_offset parameter and deal with it.
8074 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
8075 pass it to thunk_adjust. Do not call the target hook if it's non-zero
8076 or if the thunk is external or local. Fix formatting. Do not chain
8077 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
8078 if any, in the GIMPLE representation.
8079 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
8080 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
8081 (input_node): Read indirect_offset field.
8082 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
8083 call to thunk_adjust.
8084 * tree-nested.c (struct nesting_info): Add thunk_p field.
8085 (create_nesting_tree): Set it.
8086 (convert_all_function_calls): Copy static chain from targets to thunks.
8087 (finalize_nesting_tree_1): Return early for thunks.
8088 (unnest_nesting_tree_1): Do not finalize thunks.
8089 (gimplify_all_functions): Do not gimplify thunks.
8091 2018-09-28 David Malcolm <dmalcolm@redhat.com>
8093 * opt-suggestions.c (option_proposer::build_option_suggestions):
8094 Release "option_values".
8096 2018-09-28 David Malcolm <dmalcolm@redhat.com>
8098 * coverage.c (get_coverage_counts): Convert problem-reporting dump
8099 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
8100 * dumpfile.c (kind_as_string): New function.
8101 (dump_loc): Rather than a hardcoded prefix of "note: ", use
8102 kind_as_string to vary the prefix based on dump_kind.
8103 (selftest::test_capture_of_dump_calls): Update for above.
8105 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
8107 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
8108 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
8110 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
8112 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
8113 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
8114 INVALID_REGNUM instead of FPSR_REG.
8115 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
8116 * config/i386/i386.md: Update comment of FP compares.
8117 (fldenv): Do not clobber FPSR_REG.
8119 2018-09-28 Richard Biener <rguenther@suse.de>
8121 * tree.h (BLOCK_ORIGIN): New.
8122 * omp-expand.c (grid_expand_target_grid_body): Assign
8123 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
8124 * tree-inline.c (remap_block): Likewise.
8125 * auto-profile.c (get_function_decl_from_block): Simplify
8126 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
8127 * langhooks.c (lhd_print_error_function): Likewise.
8128 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
8130 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
8131 * tree.c (block_nonartificial_location): Likewise.
8132 (block_ultimate_origin): Likewise.
8133 * tree-pretty-print.c (percent_K_format): Likewise. Remove
8134 no longer needed LTO case.
8136 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
8137 Jan Hubicka <jh@suse.cz>
8138 Martin Jambor <mjambor@suse.cz>
8140 * simplify-rtx.c (simplify_merge_mask): New function.
8141 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
8142 same masks are used in op1 or op2.
8143 (test_vec_merge): New function.
8144 (test_vector_ops): Call test_vec_merge.
8146 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
8148 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
8149 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
8150 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
8151 (cypress_costs): Set it.
8152 (supersparc_costs): Likewise.
8153 (hypersparc_costs): Likewise.
8154 (leon_cost): Likewise.
8155 (leon3_costs): Likewise.
8156 (sparclet_costs): Likewise.
8157 (ultrasparc_costs): Likewise.
8158 (ultrasparc_costs): Likewise.
8159 (niagara_costs): Likewise.
8160 (niagara2_costs): Likewise.
8161 (niagara3_costs): Likewise.
8162 (niagara4_costs): Likewise.
8163 (niagara7_costs): Likewise.
8164 (m8_costs): Likewise.
8165 (TARGET_CAN_FOLLOW_JUMP): Define.
8166 (pass_work_around_errata::gate): Minor tweak.
8167 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
8168 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
8169 Automaitcally clear MASK_FSMULD mask for V7 processors.
8170 (sparc_can_follow_jump): New static function.
8171 (output_ubranch): Deal with CROSSING_JUMP_P.
8172 (sparc_use_sched_lookahead): Rewrite using switch statement.
8173 (sparc_issue_rate): Reorder.
8174 (sparc_branch_cost): New function.
8176 2018-09-27 Martin Sebor <msebor@redhat.com>
8178 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
8179 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
8180 (int_fits_type_p): Same.
8182 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
8184 * config/i386/i386.md (FPCR_REG): Remove.
8185 (UNSPEC_FLDCW): Remove.
8186 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
8187 (x86_fldcw_1): Remove insn pattern.
8188 (fnstenv): Do not clobber FPCR_REG.
8190 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
8191 (CALL_USED_REGISTERS): Ditto.
8192 (REG_ALLOC_ORDER): Ditto.
8193 (REG_CLASS_CONTENTS): Ditto.
8194 (HI_REGISTER_NAMES): Ditto.
8195 (ADDITIONAL_REGISTER_NAMES): Use defines instead
8196 of numerical constants.
8197 * config/i386/i386.c (regclass_map): Remove fpsr register.
8198 (dbx_register_map): Ditto.
8199 (dbx64_register_map): Ditto.
8200 (svr4_dbx_register_map): Ditto.
8201 (print_reg): Do not handle FPCR_REG.
8203 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
8206 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
8207 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
8208 Delete, always treat as true.
8209 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
8210 Ditto. Simplify remaining code.
8211 * config/powerpcspe/powerpcspe.h: Ditto.
8212 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
8213 Simplify remaining code.
8214 (rs6000_expand_builtin): Ditto.
8215 * config/rs6000/rs6000.h: Ditto.
8216 * configure.ac: Ditto.
8217 * configure: Regenerate.
8219 2018-09-27 Martin Liska <mliska@suse.cz>
8221 * coverage.c (get_coverage_counts): Revert the formatting
8222 of missing profile opt info.
8224 2018-09-27 Richard Biener <rguenther@suse.de>
8228 * dwarf2out.c (set_block_origin_self): Do not mark outermost
8229 block as we do not output that.
8230 (gen_inlined_subroutine_die): Elide the originally outermost
8231 block, matching what we do for concrete instances.
8232 (decls_for_scope): Add parameter specifying whether to recurse
8235 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
8236 Tom de Vries <tom@codesourcery.com>
8240 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
8241 STORE_FLAG_VALUE == 1.
8243 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
8245 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
8246 constant definitions.
8247 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
8248 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
8249 ("speculation_barrier"): New expander definition.
8251 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
8253 PR gcov-profile/86957
8254 * common.opt: New warning option -Wmissing-profile.
8255 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
8256 * doc/invoke.texi: Document -Wmissing-profile.
8258 2018-09-26 Jim Wilson <jimw@sifive.com>
8260 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
8261 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
8263 2018-09-26 Martin Sebor <msebor@redhat.com>
8265 * tree.c (zerop): Change return type to bool.
8266 (integer_zerop, integer_onep, integer_each_onep): Same.
8267 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
8268 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
8269 (real_onep, real_minus_onep, chain_index): Same.
8270 (print_type_hash_statistics, type_list_equal): Same.
8271 * tree.h (zerop): Same.
8272 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
8273 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
8274 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
8275 (real_onep, real_minus_onep, chain_index): Same.
8276 (print_type_hash_statistics, type_list_equal): Same.
8278 2018-09-26 Jim Wilson <jimw@sifive.com>
8280 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
8282 2018-09-26 Jakub Jelinek <jakub@redhat.com>
8285 * config/i386/i386.c: Include debug.h and dwarf2out.h.
8286 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
8289 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
8291 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
8293 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
8295 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
8296 and FP_SECOND_SSE_REGS.
8297 (REG_CLASS_NAMES): Ditto.
8298 (REG_CLASS_CONTENTS): Ditto.
8299 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
8300 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
8301 (ix86_preferred_output_reload_class): Ditto.
8302 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
8303 clobber constraint to "=&f".
8304 (fix_truncdi_i387): Ditto.
8305 (lrintxfdi2): Ditto.
8306 (fistdi2_<rounding>): Ditto.
8307 (fpremxf4_i387): Change "=u" constraint to "=f".
8308 (fprem1xf4_i387): Ditto.
8310 (fptanxf4_i387): Ditto.
8311 (fxtractxf3_i387): Ditto.
8312 (fscalexf4_i387): Ditto.
8313 (atan2xf3): Change "u" constraint to "f".
8314 (fyl2xxf3_i387): Ditto.
8315 (fyl2xp1xf3_i387): Ditto.
8317 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
8320 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
8321 for removed I387_MASK_PM entity.
8324 2018-09-26 Jeff Law <law@redhat.com>
8326 2018-09-26 Alexey Neyman <stilor@att.net>
8328 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
8329 headers are no longer pulled in by <isl/val.h>.
8331 2018-09-26 Richard Biener <rguenther@suse.de>
8334 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
8335 or concrete instance DIE to the tree. Create abstract origin
8336 attributes also for concrete instances.
8338 2018-09-26 Alexey Neyman <stilor@att.net>
8340 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
8341 headers are no longer pulled in by <isl/val.h>.
8343 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
8345 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
8346 Use new helper functions.
8347 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
8348 Use new helper functions.
8349 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
8350 aarch_mm_needs_release): New declarations.
8351 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
8352 aarch_mm_needs_release): New.
8354 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
8356 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
8357 (arm32_output_mi_thunk): Deal with long calls.
8359 2018-09-26 Richard Biener <rguenther@suse.de>
8363 * tree-inline.c (expand_call_inline): When the location
8364 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
8365 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
8366 the inserted BLOCK to make inlined_function_outer_scope_p
8368 * dwarf2out.c (add_call_src_coords_attributes): Do not add
8369 coords for reserved locations.
8371 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
8373 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
8374 (*call_indirect_nonlocal_sysv<mode>): Ditto.
8375 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
8376 (*sibcall_nonlocal_sysv<mode>): Ditto.
8377 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
8378 (<bd>_<mode>): Ditto.
8379 (<bd>tf_<mode>): Ditto.
8381 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
8383 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
8384 control string as a list of templates instead of as C code.
8385 (*altivec_movti): Ditto.
8386 * config/rs6000/darwin.md (movdf_low_di): Ditto.
8388 2018-09-25 Jim Wilson <jimw@sifive.com>
8390 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
8391 when target symbol is weak.
8393 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
8396 * builtins.c (unterminated_array): Simplify.
8397 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
8398 where pointer arithmetic is safe.
8400 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
8403 * config/rs6000/altivec.md (altivec_vspltb): Use
8404 const_0_to_15_operand instead of u5bit_cint_operand.
8405 (*altivec_vspltb_internal): Ditto.
8406 (altivec_vspltb_direct): Ditto.
8407 (altivec_vsplth): Use const_0_to_7_operand instead of
8409 (*altivec_vsplth_internal): Ditto.
8410 (altivec_vsplth_direct): Ditto.
8411 (altivec_vspltw): Use const_0_to_3_operand instead of
8413 (*altivec_vspltw_internal): Ditto.
8414 (altivec_vspltw_direct): Ditto.
8415 (altivec_vspltsf): Ditto.
8416 (*altivec_vspltsf_internal): Ditto.
8417 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
8418 various splats with the proper size immediate. Reorder the various
8419 cases by ascending size of immediate, and put all such together.
8421 2018-09-25 Richard Biener <rguenther@suse.de>
8424 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
8426 (maybe_create_die_with_external_ref): Do not create
8427 DW_TAG_imported_unit here.
8428 (add_abstract_origin_attribute): Handle external BLOCK refs.
8429 (dwarf2out_abstract_function): Simplify LTO case.
8430 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
8431 rather than using maybe_create_die_with_external_ref.
8433 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
8436 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
8437 (frndintxf2_mask_pm_i387): Ditto.
8438 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
8439 Enable for !flag_trapping_math.
8440 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
8441 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
8442 Change operand 1 predicate to nonimmediate_operand.
8443 (attr "i387_cw"): Remove mask_pm.
8444 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
8445 (enum ix86_entity): Remove I387_MASK_PM.
8446 * config/i386/i386.c (ix86_i387_mode_needed): Do not
8447 handle I387_MASK_PM.
8448 (ix86_mode_needed): Ditto.
8449 (ix86_mode_after): Ditto.
8450 (ix86_mode_entry): Ditto.
8451 (ix86_mode_exit): Ditto.
8452 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
8454 2018-09-25 Jakub Jelinek <jakub@redhat.com>
8456 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
8457 to_update_switch_stmts to vNULL instead of calling create on them
8460 2018-09-25 Richard Biener <rguenther@suse.de>
8462 PR tree-optimization/87402
8463 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
8464 (visit_phi): Re-instantiate handling of supposed to be VARYING
8465 but non-VARYING backedge value.
8467 2018-09-25 Richard Biener <rguenther@suse.de>
8470 * dwarf2out.c (struct sym_off_pair): New.
8471 (external_die_map): New global.
8472 (lookup_decl_die): When in LTO create DIEs lazily from the
8474 (lookup_block_die): New function, create DIEs lazily in LTO.
8475 (equate_block_to_die): New function.
8476 (dwarf2out_die_ref_for_decl): During WPA get the association
8477 from the external DIE map.
8478 (dwarf2out_register_external_die): Record mapping into the
8480 (maybe_create_die_with_external_ref): New function split out from
8481 DIE generation part of old dwarf2out_register_external_die.
8482 (add_abstract_origin_attribute): Do not return the DIE. When
8483 in LTO reference externals directly.
8484 (dwarf2out_abstract_function): When in LTO ignore calls for
8485 decls with external DIEs (already present abstract instances).
8486 (gen_call_site_die): Adjust.
8487 (add_high_low_attributes): Likewise.
8488 (gen_lexical_block_die): Likewise.
8489 (gen_inlined_subroutine_die): Likewie.
8490 (gen_block_die): Likewise.
8491 (dwarf2out_inline_entry): Likewise.
8492 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
8495 2018-09-25 Martin Liska <mliska@suse.cz>
8497 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
8498 integers and not by a float value.
8500 2018-09-25 Martin Liska <mliska@suse.cz>
8503 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
8505 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
8506 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
8507 * godump.c (go_format_type): Remove extra parenthesis.
8509 2018-09-25 Martin Liska <mliska@suse.cz>
8511 * alias.c (set_dest_equal_p): Remove unused function.
8512 * config/i386/i386.c (def_builtin_pure2): Likewise.
8513 * diagnostic-show-locus.c (class layout): Remove
8515 (layout::layout): Likewise here.
8516 * dump-context.h (class temp_dump_context): Likewise.
8517 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
8518 (add_AT_loclistsptr): Likewise.
8519 (add_AT_offset): Likewise.
8520 (get_AT_hi_pc): Likewise.
8521 (is_comdat_die): Likewise.
8522 (type_is_enum): Likewise.
8523 (ceiling): Likewise.
8524 (add_AT_vms_delta): Likewise.
8525 (is_class_die): Likewise.
8526 * edit-context.c (class line_event): Remove unused field.
8527 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
8529 * ipa-cp.c (ipa_get_vr_lat): Likewise.
8530 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
8531 (ok_for_base_p_nonstrict): Likewise.
8532 * tree-chrec.c (is_not_constant_evolution): Likewise.
8533 (chrec_fold_poly_cst): Likewise.
8534 * tree-if-conv.c (has_pred_critical_p): Likewise.
8535 * tree-ssa-coalesce.c (print_exprs): Likewise.
8536 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
8537 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
8538 * tree-vrp.c (value_ranges_intersect_p): Likewise.
8539 (value_range_nonnegative_p): Likewise.
8541 2018-09-25 Martin Liska <mliska@suse.cz>
8543 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
8544 Do not handle "GNU Pascal".
8545 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
8547 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
8548 from documentation. Likewise.
8549 * dbxout.c (dbxout_range_type): Likewise.
8550 * doc/cpp.texi: Likewise.
8551 * doc/extend.texi: Likewise.
8552 * doc/frontends.texi: Likewise.
8553 * doc/invoke.texi: Remove Pascal entry.
8554 * tree.def (CLEANUP_POINT_EXPR): Likewise.
8555 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
8557 2018-09-25 Martin Liska <mliska@suse.cz>
8560 * doc/invoke.texi: Document all parameters and remove default
8563 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
8566 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
8567 contains HOST_WIDE_INTs when computing its size.
8569 2018-09-24 Jim Wilson <jimw@sifive.com>
8572 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
8574 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
8576 2018-09-24 Andrew Pinski <apinski@marvell.com>
8578 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
8579 access prev before checking it for NULLness in the
8580 AARCH64_FUSE_CMP_BRANCH case.
8582 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
8585 * config/i386/i386.c (rest_of_insert_endbranch): Set
8586 endbr_queued_at_entrance to true and don't insert ENDBR if
8587 x86_function_profiler will be called.
8588 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
8590 * config/i386/i386.h (machine_function): Add
8591 endbr_queued_at_entrance.
8593 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
8595 * genattrtab.c (mk_attr_alt): Use alternative_mask.
8596 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
8598 (check_attr_test): Use alternative_mask.
8599 (get_attr_value): Likewise.
8600 (compute_alternative_mask): Use alternative_mask and XWINT.
8601 (make_alternative_compare): Use alternative_mask.
8602 (attr_alt_subset_p): Use XWINT.
8603 (attr_alt_subset_of_compl_p): Likewise.
8604 (attr_alt_intersection): Use alternative_mask and XWINT.
8605 (attr_alt_union): Likewise.
8606 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
8607 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
8608 (simplify_test_exp): Use alternative_mask and XWINT.
8609 (write_test_expr): Use alternative_mask and XWINT, adjust bit
8610 number calculation to support 64 bits. Generate code that
8611 checks 64-bit masks.
8612 (main): Use alternative_mask.
8613 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
8615 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
8618 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
8619 RETURN+USE when returning via %r14.
8621 2018-09-24 Martin Liska <mliska@suse.cz>
8623 * gcov.c (output_lines): Print colorization legend
8624 for both flag_use_colors and flag_use_hotness_colors.
8627 2018-09-24 Martin Liska <mliska@suse.cz>
8629 * coverage.c (get_coverage_counts): Use warning_at
8630 with current_function_decl location. Use %qD in warning
8633 2018-09-24 Martin Liska <mliska@suse.cz>
8635 * memory-block.h (memory_block_pool::release): Annotate with
8636 valgrind that the memory is not accessible.
8638 2018-09-24 Martin Liska <mliska@suse.cz>
8641 * asan.c: Make asan_handled_variables extern.
8643 * cfgexpand.c (expand_stack_vars): Make sure
8644 a representative is unpoison if another
8645 variable in the partition is handled by
8646 use-after-scope sanitization.
8648 2018-09-24 Richard Biener <rguenther@suse.de>
8650 PR tree-optimization/63155
8651 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
8652 the worklist when the edge of the respective argument isn't
8655 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
8657 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
8658 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
8659 (MASK_CLASS_P): Update for rename.
8660 (MAYBE_MASK_CLASS_P): Ditto.
8661 (REG_CLASS_NAMES): Update.
8662 (REG_CLASS_CONTENT): Update.
8663 * config/i386/i386.c (regclass_map): Update for MASK_REG
8664 and ALL_MASK_REGS rename.
8665 * config/i386/constraints.md (Yk): Update for rename.
8668 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
8670 * config/i386/i386.h (enum reg_class): Remove
8671 EVEX_SSE_REGS and MOD4_SSE_REGS.
8672 (REG_CLASS_NAMES): Update.
8673 (REG_CLASS_CONTENT): Update.
8674 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
8675 registers as ALL_SSE_REGS.
8676 (ix86_additional_allocno_class_p): Remove.
8677 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
8678 (ix86_register_priority): Lower priority of EVEX SSE registers.
8679 Use IN_RANGE macro where appropriate.
8680 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
8681 AVX-5124VNNIW checks.
8682 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
8683 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
8684 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
8685 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
8686 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
8687 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
8688 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
8689 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
8690 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
8691 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
8692 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
8693 * config/i386/constraints.md (Yh): Remove.
8695 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
8697 * config/i386/i386.c (regclass_map): Declare integer REX registers
8700 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
8702 * doc/service.texi (Service): Switch the fsf.org link to https.
8704 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
8707 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
8708 Define to speculation_safe_value_not_needed.
8710 2018-09-21 Florian Weimer <fweimer@redhat.com>
8713 * doc/extend.texi (Common Function Attributes): Mention that
8714 noreturn suppresses tail call optimization.
8716 2018-09-21 Jeff Law <law@redhat.com>
8718 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
8719 vr_values::cleanup_edges_and_switches.
8720 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
8722 (identify_jump_threads): Remove EDGE_IGNORE handling.
8723 (execute_vrp): Move handling of to_remove_edges and
8724 to_update_switch_stmts into vr_values class member functions.
8725 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
8726 (to_update_switch_stmts): Likewise.
8727 * vr-values.c: Include cfghooks.h.
8728 (vr_values::vr_values): Initialize to_remove_edges and
8729 to_update_switch_stmts.
8730 (vr_values::~vr_values): Verify to_remove_edges and
8731 to_update_switch_stmts are empty.
8732 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
8733 (vr_values::cleanup_edges_and_switches): New member function.
8734 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
8735 function. Add new data members.
8737 2018-09-21 David Malcolm <dmalcolm@redhat.com>
8739 PR tree-optimization/87309
8740 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
8741 calls with pflags and alt_flags.
8742 (selftest::test_capture_of_dump_calls): Add test of interaction of
8743 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
8745 2018-09-21 Olivier Hainque <hainque@adacore.com>
8747 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
8749 2018-09-21 Olivier Hainque <hainque@adacore.com>
8751 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
8753 2018-09-21 Olivier Hainque <hainque@adacore.com>
8755 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
8756 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
8757 (VXWORKS_LIBS_RTP): Minor reordering.
8759 2018-09-21 Olivier Hainque <hainque@adacore.com>
8761 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
8762 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
8764 2018-09-21 Olivier Hainque <hainque@adacore.com>
8766 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
8767 (PTRDIFF_TYPE): Likewise.
8769 2018-09-21 Olivier Hainque <hainque@adacore.com>
8771 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
8772 triplet, similar to support for VxWorks7.
8773 * config/vxworks-dummy.h: Provide a default definition
8774 of TARGET_VXWORKS64 to 0.
8776 2018-09-21 Olivier Hainque <hainque@adacore.com>
8778 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
8779 * config/vxworks-dummy.h: here.
8781 2018-09-21 Olivier Hainque <hainque@adacore.com>
8783 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
8785 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
8786 Bo Zhou <zbo.zhou@hisilicon.com>
8788 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
8789 * config/aarch64/aarch64-tune.md: Regenerated.
8790 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
8791 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
8792 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
8794 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
8795 Julian Brown <julian@codesourcery.com>
8797 * builtins.c (get_builtin_sync_mem): Handle address spaces.
8799 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
8801 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
8802 if the call takes a static chain.
8804 2018-09-21 Martin Liska <mliska@suse.cz>
8806 * auto-profile.c (autofdo_source_profile::read): Do not
8808 (read_profile): Do not add working sets.
8809 (read_autofdo_file): Remove sum_all.
8810 (afdo_callsite_hot_enough_for_early_inline): Remove const
8812 * coverage.c (struct counts_entry): Remove gcov_summary.
8813 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
8814 do not support GCOV_TAG_PROGRAM_SUMMARY.
8815 (get_coverage_counts): Remove summary and expected
8817 * coverage.h (get_coverage_counts): Likewise.
8818 * doc/gcov-dump.texi: Remove -w option.
8819 * gcov-dump.c (dump_working_sets): Remove.
8820 (main): Do not support '-w' option.
8821 (print_usage): Likewise.
8822 (tag_summary): Likewise.
8823 * gcov-io.c (gcov_write_summary): Do not dump
8825 (gcov_read_summary): Likewise.
8826 (gcov_histo_index): Remove.
8827 (gcov_histogram_merge): Likewise.
8828 (compute_working_sets): Likewise.
8829 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
8831 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
8832 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
8833 (GCOV_HISTOGRAM_SIZE): Remove.
8834 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
8835 (struct gcov_summary): Simplify rapidly just
8836 to runs and sum_max fields.
8837 (gcov_histo_index): Remove.
8838 (NUM_GCOV_WORKING_SETS): Likewise.
8839 (compute_working_sets): Likewise.
8840 * gcov-tool.c (print_overlap_usage_message): Remove
8841 trailing empty line.
8842 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
8843 (output_lines): Remove program related line.
8844 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
8845 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
8847 (input_profile_summary): Do not read it.
8848 (merge_profile_summaries): And do not merge it.
8849 (input_symtab): Do not call removed function.
8850 * modulo-sched.c (sms_schedule): Do not print sum_max.
8851 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
8852 removed when histogram method was invented.
8853 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
8855 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
8857 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
8858 and dump selected value.
8859 * profile.c (add_working_set): Remove.
8860 (get_working_sets): Likewise.
8861 (find_working_set): Likewise.
8862 (get_exec_counts): Do not work with working sets.
8863 (read_profile_edge_counts): Do not inform as sum_max is removed.
8864 (compute_branch_probabilities): Likewise.
8865 (compute_value_histograms): Remove argument for call of
8866 get_coverage_counts.
8867 * profile.h: Do not make gcov_summary const.
8869 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
8871 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
8873 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
8875 PR tree-optimization/86990
8876 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
8877 Check that the entire merged store group is made of constants only for
8880 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
8882 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
8883 (VTABLE_VERIFICATION_SPEC): Likewise.
8884 (SANITIZER_EARLY_SPEC): Likewise.
8885 (SANITIZER_SPEC): Likewise.
8886 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
8887 * doc/invoke.texi (Link Options): Document -r.
8889 2018-09-20 Richard Biener <rguenther@suse.de>
8892 * gimplify.c (gimplify_expr): Retain alignment of
8893 addressable lvalue in dereference.
8895 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
8898 * configure.ac: Check for .loc is_stmt support.
8899 * configure, config.in: Rebuilt.
8900 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
8903 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8905 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
8906 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
8909 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8911 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
8913 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
8916 2018-09-20 Marek Polacek <polacek@redhat.com>
8918 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
8920 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
8922 PR tree-optimization/87288
8923 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
8924 into account when determining PEELING_FOR_NITERS.
8926 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
8928 PR tree-optimization/86877
8929 * tree-vect-loop.c (vect_analyze_loop_2): Call
8930 vect_verify_datarefs_alignment.
8932 2018-09-19 Marek Polacek <polacek@redhat.com>
8934 * doc/invoke.texi: Document -Wclass-conversion.
8936 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
8938 * config/pa/pa.c (pa_adjust_priority): Delete.
8939 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
8941 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
8942 (atomic_storehi): Likewise.
8943 (atomic_storesi): Likewise.
8944 (atomic_loaddi): Restore compare and swap exchange loop code.
8946 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8948 PR rtl-optimization/86902
8949 * combine.c (try_combine): When changing the CC mode used, don't change
8950 an unrelated mode in other_insn to that new CC mode.
8952 2018-09-19 David Malcolm <dmalcolm@redhat.com>
8954 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
8955 with %T in place of calls to dump_generic_expr.
8956 (prune_runtime_alias_test_list): Likewise.
8957 (create_runtime_alias_checks): Likewise.
8958 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
8959 (vect_analyze_data_ref_dependence): Likewise.
8960 (vect_slp_analyze_data_ref_dependence): Likewise.
8961 (vect_record_base_alignment): Likewise. Use %G in place of call
8962 to dump_gimple_stmt.
8963 (vect_compute_data_ref_alignment): Likewise.
8964 (verify_data_ref_alignment): Likewise.
8965 (vect_find_same_alignment_drs): Likewise.
8966 (vect_analyze_group_access_1): Likewise.
8967 (vect_analyze_data_ref_accesses): Likewise.
8968 (dependence_distance_ge_vf): Likewise.
8969 (dump_lower_bound): Likewise.
8970 (vect_prune_runtime_alias_test_list): Likewise.
8971 (vect_find_stmt_data_reference): Likewise.
8972 (vect_analyze_data_refs): Likewise.
8973 (vect_create_addr_base_for_vector_ref): Likewise.
8974 (vect_create_data_ref_ptr): Likewise.
8975 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
8976 (vect_can_advance_ivs_p): Likewise.
8977 (vect_update_ivs_after_vectorizer): Likewise.
8978 (vect_gen_prolog_loop_niters): Likewise.
8979 (vect_prepare_for_masked_peels): Likewise.
8980 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
8981 (vect_determine_vectorization_factor): Likewise.
8982 (vect_is_simple_iv_evolution): Likewise.
8983 (vect_analyze_scalar_cycles_1): Likewise.
8984 (vect_analyze_loop_operations): Likewise.
8985 (report_vect_op): Likewise.
8986 (vect_is_slp_reduction): Likewise.
8987 (check_reduction_path): Likewise.
8988 (vect_is_simple_reduction): Likewise.
8989 (vect_create_epilog_for_reduction): Likewise.
8990 (vect_finalize_reduction:): Likewise.
8991 (vectorizable_induction): Likewise.
8992 (vect_transform_loop_stmt): Likewise.
8993 (vect_transform_loop): Likewise.
8994 (optimize_mask_stores): Likewise.
8995 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
8996 (vect_split_statement): Likewise.
8997 (vect_recog_over_widening_pattern): Likewise.
8998 (vect_recog_average_pattern): Likewise.
8999 (vect_determine_min_output_precision_1): Likewise.
9000 (vect_determine_precisions_from_range): Likewise.
9001 (vect_determine_precisions_from_users): Likewise.
9002 (vect_mark_pattern_stmts): Likewise.
9003 (vect_pattern_recog_1): Likewise.
9004 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
9005 (vect_record_max_nunits): Likewise.
9006 (vect_build_slp_tree_1): Likewise.
9007 (vect_build_slp_tree_2): Likewise.
9008 (vect_print_slp_tree): Likewise.
9009 (vect_analyze_slp_instance): Likewise.
9010 (vect_detect_hybrid_slp_stmts): Likewise.
9011 (vect_detect_hybrid_slp_1): Likewise.
9012 (vect_slp_analyze_operations): Likewise.
9013 (vect_slp_analyze_bb_1): Likewise.
9014 (vect_transform_slp_perm_load): Likewise.
9015 (vect_schedule_slp_instance): Likewise.
9016 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
9017 (vect_mark_stmts_to_be_vectorized): Likewise.
9018 (vect_init_vector_1): Likewise.
9019 (vect_get_vec_def_for_operand): Likewise.
9020 (vect_finish_stmt_generation_1): Likewise.
9021 (vect_check_load_store_mask): Likewise.
9022 (vectorizable_call): Likewise.
9023 (vectorizable_conversion): Likewise.
9024 (vectorizable_operation): Likewise.
9025 (vectorizable_load): Likewise.
9026 (vect_analyze_stmt): Likewise.
9027 (vect_is_simple_use): Likewise.
9028 (vect_get_vector_types_for_stmt): Likewise.
9029 (vect_get_mask_type_for_stmt): Likewise.
9030 * tree-vectorizer.c (increase_alignment): Likewise.
9032 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
9034 * doc/rtl.texi: Adjust vec_select description.
9035 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
9036 non-constant selectors.
9038 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
9040 * config/aarch64/aarch64-protos.h
9041 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
9042 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
9043 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
9044 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
9045 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
9046 (AARCH64_FL_PROFILE): Move index so flags are ordered.
9047 (AARCH64_ISA_RCPC8_4): New flag.
9048 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
9049 to aarch64_offset_9bit_signed_unscaled_p.
9050 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
9052 * config/aarch64/constraints.md (Ust): New constraint.
9053 * config/aarch64/predicates.md.
9054 (aarch64_9bit_offset_memory_operand): New predicate.
9055 (aarch64_rcpc_memory_operand): New predicate.
9057 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
9059 PR rtl-optimization/87361
9060 * rtlanal.c (nonzero_bits1): Revert accidental change.
9062 2018-09-19 Richard Biener <rguenther@suse.de>
9064 PR tree-optimization/87349
9065 PR tree-optimization/87342
9066 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
9068 2018-09-18 Marek Polacek <polacek@redhat.com>
9070 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
9071 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
9073 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
9075 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
9077 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
9079 PR rtl-optimization/86882
9080 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
9082 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
9084 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
9085 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
9087 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
9090 * doc/invoke.texi (Link Options): Fix formatting and grammar.
9092 2018-09-18 Richard Biener <rguenther@suse.de>
9095 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
9096 (compute_samebase_partition_bases): Likewise.
9097 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
9098 (gimple_can_coalesce_p): Simplify.
9100 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
9102 Handle a library implementation of ffs calling __builtin_ffs.
9103 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
9104 (mmix_init_libfuncs): New function: make __builtin_ffs expand
9107 2018-09-17 David Malcolm <dmalcolm@redhat.com>
9109 * diagnostic-show-locus.c (class layout_range): Add field
9111 (layout_range::layout_range): Add "original_idx" param and use it
9112 to initialize new field.
9113 (make_range): Use 0 for original_idx.
9114 (layout::layout): Pass in index to calls to
9115 maybe_add_location_range.
9116 (layout::maybe_add_location_range): Add param "original_idx" and
9117 pass it on to layout_range.
9118 (layout::print_any_labels): Pass on range->m_original_idx to
9120 (gcc_rich_location::add_location_if_nearby): Use 0 for
9122 * gcc-rich-location.h (text_range_label::get_text): Update for new
9124 (range_label_for_type_mismatch::get_text): Likewise.
9126 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
9128 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
9130 2018-09-17 David Malcolm <dmalcolm@redhat.com>
9132 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
9133 format_string_diagnostic_t.
9134 (fmtwarn_n): Likewise.
9135 * substring-locations.c
9136 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
9137 (format_warning_n_va): Convert to...
9138 (format_string_diagnostic_t::emit_warning_n_va): ...this.
9139 (format_warning_va): Convert to...
9140 (format_string_diagnostic_t::emit_warning_va): ...this.
9141 (format_warning_at_substring): Convert to...
9142 (format_string_diagnostic_t::emit_warning): ...this.
9143 (format_warning_at_substring_n): Convert to...
9144 (format_string_diagnostic_t::emit_warning_n): ...this.
9145 * substring-locations.h (class format_string_diagnostic_t): New
9147 (format_warning_va): Convert to
9148 format_string_diagnostic_t::emit_warning_va.
9149 (format_warning_n_va): Convert to
9150 format_string_diagnostic_t::emit_warning_n_va.
9151 (format_warning_at_substring): Convert to
9152 format_string_diagnostic_t::emit_warning.
9153 (format_warning_at_substring_n): Convert to
9154 format_string_diagnostic_t::emit_warning_n.
9156 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
9157 Bernd Schmidt <bernds_cb1@t-online.de>
9159 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
9162 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
9164 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
9165 operand 0 predicate to nonimmediate operand.
9166 (rint<mode>2_frndint): Remove insn pattern.
9167 (rint<mode>2): Change operand 1 predicate to general_operand.
9168 Extend operand 1 to XFmode and generate rintxf2 insn.
9169 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
9170 Do not use X87MODEF mode macro.
9171 (frndintxf2_<rounding>_i387): Rename from
9172 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
9173 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
9174 to XFmode and generate significandxf3 insn.
9176 2018-09-17 Richard Biener <rguenther@suse.de>
9178 PR tree-optimization/87328
9179 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
9180 visiting unexecutable backedges when not iterating.
9181 (do_rpo_vn): Mark all edges not executable even when not
9184 2018-09-17 Martin Jambor <mjambor@suse.cz>
9187 * doc/invoke.texi (Warning Options): Likewise.
9189 2018-09-17 Richard Biener <rguenther@suse.de>
9191 PR tree-optimization/87301
9192 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
9193 clean EH info from leftover copy assignments.
9195 2018-09-17 Martin Liska <mliska@suse.cz>
9197 PR gcov-profile/85871
9198 * gcov.c (output_intermediate_file): Fix out of bounds
9201 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
9203 * config/arc/arc.c: Object attributes for core4 not reflected
9205 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
9208 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
9210 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
9212 2018-09-17 Martin Liska <mliska@suse.cz>
9214 * doc/gcov.texi: Document new option --use-hotness-colors.
9215 * gcov.c (struct source_info): Declare new field.
9216 (source_info::source_info): Set default for maximum_count.
9217 (print_usage): Add new -q option.
9218 (process_args): Process it.
9219 (accumulate_line_info): Save src->maximum_count.
9220 (output_line_beginning): Make color line number if
9221 flag_use_hotness_colors is set.
9222 (output_line_details): Pass default argument value.
9223 (output_lines): Pass src->maximum_count.
9225 2018-09-17 Martin Liska <mliska@suse.cz>
9227 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
9228 Use processor_names table.
9229 * config/i386/i386.c (ix86_default_align): Use
9230 processor_cost_table for alignment values.
9231 (ix86_option_override_internal): Use processor_names.
9232 (ix86_function_specific_print): Likewise.
9233 * config/i386/i386.h (struct processor_costs):
9234 Add alignment values.
9235 (struct ptt): Remove and replace with const char *.
9236 * config/i386/x86-tune-costs.h (struct processor_costs):
9237 Declare default alignments for all costs.
9239 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
9241 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
9242 symbolics or VR_VARYING ranges for ABS_EXPR.
9243 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
9244 when range will wrap.
9246 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
9249 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
9250 before and after a JUMP_TABLE_DATA.
9252 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
9255 * dojump.c (do_compare_and_jump): Canonicalize function pointers
9256 when one operand is a function pointer. Use POINTER_TYPE_P and
9257 FUNC_OR_METHOD_TYPE_P.
9258 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
9259 * fold-const.c (build_range_check): Likewise.
9260 * match.pd (simple_comparison): Likewise.
9262 2018-09-14 David Malcolm <dmalcolm@redhat.com>
9265 * spellcheck.c (get_edit_distance_cutoff): New function.
9266 (selftest::test_edit_distance_unit_test_oneway): Rename to...
9267 (selftest::test_get_edit_distance_one_way): ...this.
9268 (selftest::test_get_edit_distance_unit): Rename to...
9269 (selftest::test_get_edit_distance_both_ways): ...this.
9270 (selftest::test_edit_distances): Move tests to this new function,
9271 and test some more pairs of strings. Update for above renaming.
9272 (selftest::get_old_cutoff): New function.
9273 (selftest::test_get_edit_distance_cutoff): New function.
9274 (selftest::assert_suggested_for): New function.
9275 (ASSERT_SUGGESTED_FOR): New macro.
9276 (selftest::assert_not_suggested_for): New function.
9277 (ASSERT_NOT_SUGGESTED_FOR): New macro.
9278 (selftest::test_suggestions): New function.
9279 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
9280 tests to selftest::test_edit_distances and call it. Add calls to
9281 selftest::test_get_edit_distance_cutoff and
9282 selftest::test_suggestions.
9283 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
9284 (best_match::consider): Replace hard-coded cutoff calculation with
9286 (best_match::get_cutoff): New declaration.
9287 (best_match::get_best_meaningful_candidate): Likewise.
9289 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
9291 * builtins.c (fold_builtin_strlen): Remove TODO comment.
9293 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
9296 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
9298 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
9299 terminated string literal.
9301 2018-09-14 Martin Sebor <msebor@redhat.com>
9303 * builtins.c (unterminated_array): Handle ARRAY_REF.
9304 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
9305 * builtins.h (unterminated_array): Declare extern.
9306 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
9308 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
9311 2018-09-14 Martin Sebor <msebor@redhat.com>
9312 Jeff Law <law@redhat.com>
9314 * builtins.c (unterminated_array): New.
9315 (expand_builtin_strcpy): Adjust.
9316 (expand_builtin_strcpy_args): Detect unterminated arrays.
9317 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
9318 unterminated arrays.
9319 * gimple-fold.h (get_maxval_strlen): Add argument.
9320 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
9322 * gimple-fold.c (get_range_strlen): Add argument.
9323 (get_maxval_strlen): Adjust.
9324 * gimple-fold.h (get_range_strlen): Add argument.
9326 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
9328 * config/i386/movdirintrin.h: Fix copyright year.
9330 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
9332 * reg-stack.c: Include regs.h.
9333 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
9334 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
9335 FIRST_STACK_REG, not DFmode.
9336 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
9337 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
9338 (change stack): Default register mode to the reg_raw_mode of
9339 FIRST_STACK_REG, not DFmode.
9340 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
9341 (*swapxf): Rename from swapxf.
9343 2018-09-14 Carl Love <cel@us.ibm.com>
9345 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
9346 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
9348 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
9351 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
9354 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
9357 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
9359 2018-09-14 Jason Merrill <jason@redhat.com>
9361 Fix --enable-gather-detailed-mem-stats.
9362 * hash-table.c (hash_table_usage): Change from variable to function.
9363 * hash-table.h: Adjust.
9364 * Makefile.in: Add missing dependencies on hash-table.h.
9366 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9368 PR tree-optimization/87259
9370 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
9371 execute_cse_reciprocals_1 has tried transforming.
9373 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
9375 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
9376 VR_VARYING for PLUS/MINUS_EXPR.
9378 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
9380 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
9383 2018-09-14 Richard Biener <rguenther@suse.de>
9386 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
9387 bits for the merged partition.
9389 2018-09-13 Martin Sebor <msebor@redhat.com>
9390 Bernd Edlinger <bernd.edlinger@hotmail.de>
9392 * builtins.h (c_srlen): Add argument.
9393 * builtins.c (warn_string_no_nul): New function.
9394 (c_strlen): Add argument and use it. Update recursive calls.
9395 Pass DECL argument to string_constant to get info on non
9396 terminated strings. Update *NONSTR as needed.
9397 (fold_builtin_strlen): Add argument to calls to c_strlen.
9398 Warn for unterminated arrays.
9399 (warn_string_no_null): Add prototype.
9400 * expr.c (string_constant): Update arguments. Update recursive
9401 calls appropriately. Detect missing NUL terminator and outermost
9402 declaration its missing in.
9403 Improve checks for arrays with nonzero lower bound or elements
9404 that are not a single byte. Simplify offset computation.
9405 Simplify checks for non-NUL terminated strings.
9406 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
9407 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
9409 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
9411 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
9413 * fold-const.c (c_getstr): Fix function comment. Remove unused third
9414 argument. Fix range checks.
9415 * fold-const.h (c_getstr): Adjust protoype.
9416 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
9417 string is constant but contains no NUL byte.
9419 * expr.c (string_constant): Adjust function comment.
9420 Remove bogus check for zero termination.
9422 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
9424 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
9425 (get_constant_size): Don't make STRING_CSTs larger than they are.
9426 (check_string_literal): New check function for STRING_CSTs.
9427 (output_constant): Use it.
9429 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
9432 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
9434 2018-09-13 Richard Biener <rguenther@suse.de>
9436 PR tree-optimization/87263
9437 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
9438 (struct unwind_state): Add max_rpo field.
9439 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
9440 Compute max_rpo, the max RPO number a block can be backwards reached
9441 from. Re-write non-iterating mode to a RPO ordered worklist approach,
9442 separating it from the iterating mode.
9444 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
9446 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
9447 (rfs_decision): New scheduling decision.
9449 2018-09-13 Richard Biener <rguenther@suse.de>
9452 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
9453 (vn_nary_op_insert_pieces_predicated): Do not write useless
9454 valid_dominated_by_p entry outside of the allocated storage.
9456 2018-09-13 Omar Sandoval <osandov@osandov.com>
9457 Tom de Vries <tdevries@suse.de>
9460 * dwarf2out.c (is_c): New function.
9461 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
9463 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
9466 * config/aarch64/aarch64.md (*aarch64_bfxil):
9468 * config/aarch64/constraints.md (Ulc): Define.
9469 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
9471 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
9474 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
9476 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9477 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
9478 aarch64_layout_frame call.
9479 (aarch64_expand_epilogue): Likewise.
9480 (aarch64_initial_elimination_offset): Likewise.
9481 (aarch64_get_separate_components): Likewise.
9482 (aarch64_use_return_insn_p): Likewise.
9483 (aarch64_layout_frame): Remove unneeded check.
9485 2018-09-13 Jakub Jelinek <jakub@redhat.com>
9487 * configure.ac: Only append
9488 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
9489 gcc_config_arguments if it was never reconfigured or last reconfigure
9490 was with different arguments.
9491 * configure: Regenerated.
9493 2018-09-13 Jakub Jelinek <jakub@redhat.com>
9494 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9497 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
9498 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
9500 2018-09-13 Jakub Jelinek <jakub@redhat.com>
9502 PR tree-optimization/87287
9503 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
9504 X % C == 0 to X % (unsigned) C == 0 optimization to ...
9505 * match.pd (X % C == 0): ... here. New optimization.
9507 2018-09-12 Jakub Jelinek <jakub@redhat.com>
9510 * expr.h (maybe_optimize_mod_cmp): Declare.
9511 * expr.c (mod_inv): New function.
9512 (maybe_optimize_mod_cmp): New function.
9513 (do_store_flag): Use it.
9514 * cfgexpand.c (expand_gimple_cond): Likewise.
9516 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
9517 Julian Brown <julian@codesourcery.com>
9520 * gimplify.c (gimplify_scan_omp_clauses): Set
9521 target_firstprivatize_array_bases in OpenACC parallel and kernels
9522 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
9523 OpenACC data regions.
9525 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
9527 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
9528 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
9530 2018-09-12 Richard Biener <rguenther@suse.de>
9532 PR tree-optimization/87280
9533 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
9534 edge but unreachable target.
9535 (do_rpo_vn): For conservatively handling a PHI only mark
9536 the backedge executable but not the block reachable.
9538 2018-09-12 Richard Biener <rguenther@suse.de>
9540 PR tree-optimization/87266
9541 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
9544 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
9546 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
9548 ("trunc<BFP:mode><DFP_ALL:mode>2")
9549 ("trunc<DFP_ALL:mode><BFP:mode>2")
9550 ("extend<BFP:mode><DFP_ALL:mode>2")
9551 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
9552 according to the target operand type.
9554 2018-09-12 Jakub Jelinek <jakub@redhat.com>
9555 Andreas Krebbel <krebbel@linux.ibm.com>
9557 PR tree-optimization/86844
9558 * gimple-ssa-store-merging.c
9559 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
9560 there are any overlapping stores in between them, make sure they are
9561 also coalesced or we give up completely.
9563 2018-09-12 Jakub Jelinek <jakub@redhat.com>
9566 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
9567 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
9569 2018-09-12 Tom de Vries <tdevries@suse.de>
9571 * common.opt (gdescribe-dies): Add option.
9572 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
9573 attribute for artifical and nameless decls.
9574 (dwarf2out_register_external_die): Add description attribute to
9575 external reference die.
9576 (add_desc_attribute): New functions.
9577 (gen_subprogram_die): Add description attribute to
9578 DW_TAG_call_site_parameter.
9579 * tree-pretty-print.c (print_generic_expr_to_str): New function.
9580 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
9581 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
9583 (@item -gdescribe-dies): Add.
9585 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
9587 * tree-vrp.c (vrp_shift_undefined_p): Remove.
9588 (extract_range_from_binary_expr_1: Call
9589 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
9590 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
9593 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
9595 * gimple-ssa-warn-alloca.c
9596 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
9597 field for ALLOCA_BOUND_*_LARGE.
9599 2018-09-11 Nathan Sidwell <nathan@acm.org>
9601 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
9603 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
9605 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
9606 for clobbers. Remove obsolete comment.
9608 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
9610 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
9611 mpxchk, mpxld and mpxst types.
9612 (define_attr length_immediate): Remove all processing of mpx types.
9613 (define_attr prefix_0f): Ditto.
9614 (define_attr memory): Ditto.
9616 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
9618 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
9619 (log<mode>2): Change operand 1 predicate to general_operand.
9620 Extend operand 1 to XFmode and generate logxf3 insn.
9621 (log10<mode>2): Change operand 1 predicate to general_operand.
9622 Extend operand 1 to XFmode and generate log10xf3 insn.
9623 (log2<mode>2): Change operand 1 predicate to general_operand.
9624 Extend operand 1 to XFmode and generate log2xf3 insn.
9625 (fyl2xp1_extend<mode>xf3_i387): Remove.
9626 (log1p<mode>2): Change operand 1 predicate to general_operand.
9627 Extend operand 1 to XFmode and generate log1pxf3 insn.
9628 (fxtract_extend<mode>xf3_i387): Remove.
9629 (logb<mode>2): Change operand 1 predicate to general_operand.
9630 Extend operand 1 to XFmode and generate logbxf3 insn.
9631 (ilogb<mode>2): Change operand 1 predicate to general_operand.
9632 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
9633 (significand<mode>2): Change operand 1 predicate to general_operand.
9634 Extend operand 1 to XFmode and generate significandxf3 insn.
9636 2018-09-11 Nathan Sidwell <nathan@acm.org>
9638 * gcc.c (perror_with_name, pfatal_with_name): Delete.
9639 (load_specs): Use fatal_error.
9640 (DELETE_IF_ORDINARY, process_command): Use error.
9641 (execute, run_attempt): Use fatal_error.
9643 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
9645 * diagnostic-core.h (sorry_at): New prototype.
9646 * diagnostic.c (sorry_at): New function.
9648 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
9650 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
9651 by zero as VR_UNDEFINED.
9653 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
9655 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
9656 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
9657 (<sincos>mode2): New expander.
9658 (sincos_extend<mode>xf3_i387): Remove insn pattern.
9659 (sincos -> sin, cos splitters): Remove splitter patterns.
9660 (sincos<mode>3): Change operand 2 predicate to general_operand.
9661 Extend operand 2 to XFmode and generate sincosxf3 insn.
9662 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
9663 Change operand 3 predicate to const1_operand.
9664 (fptan_extend<mode>xf4_i387): Remove insn pattern.
9665 (tanxf2): Update operands in the call to fptanxf4_i387.
9666 (tan<mode>2): Change operand 1 predicate to general_operand.
9667 Extend operand 1 to XFmode and generate tanxf3 insn.
9668 (atan2xf3): Rename from *fpatanxf3_i387.
9669 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
9670 (atan2xf3): Remove expander.
9671 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
9672 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
9673 (atan<mode>2): Change operand 1 predicate to general_operand.
9674 Extend operand 1 to XFmode and generate atanxf3 insn.
9676 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
9678 * config/i386/i386.md (x87/SSE constant load splitter): Use
9679 memory_operand instead of nonimmediate_operand for input operand
9682 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
9684 * config/i386/i386.md (float partial SSE register stall splitter): Move
9685 splitter near its instruction pattern.
9686 (float_extend partial SSE register stall splitter): Ditto.
9687 (float_truncate partial SSE register stall splitter): Ditto.
9689 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
9692 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
9693 to speculation_safe_value_not_needed.
9696 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
9697 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
9698 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
9699 leaf_function_p, instead use has_hard_reg_initial_val.
9701 2018-09-09 Nathan Sidwell <nathan@acm.org>
9703 * gcc.h (pfatal_with_name): Don't declare here.
9704 * gcc.c (pfatal_with_name): Make static.
9706 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
9708 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
9711 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
9713 PR rtl-optimization/85458
9714 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
9715 priority hook to reduce the priority of EXPR.
9717 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
9719 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
9720 DImode for x87 on 32bit targets. Conditionally disable x87 modes
9721 with X87_ENABLE_FLOAT. Remove preparation code.
9722 (*float<SWI48:mode><MODEF:mode>2): Rename from
9723 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
9724 math using "enabled" attribute.
9725 (*floatdi<MODEF:mode>2_i387): Rename from
9726 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
9727 enable for 32bit targets only.
9728 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
9730 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
9731 as operand 1 predicate. Rewrite as define_insn_and_split.
9732 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
9734 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
9736 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
9737 to fallthru to FLOAT case.
9739 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
9742 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
9743 around folding of vec_sl to handle out of range shift values.
9745 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
9747 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
9748 Update callers to gen_fix_trunc<mode>_i387_fisttp
9749 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
9750 nonimmediate_operand.
9751 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
9752 and corresponding splitters.
9753 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
9754 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
9755 (fix_truncdi_i387_with_temp): Remove insn pattern
9756 and corresponding splitters.
9757 (fix_trunc<mode>_i387): Change operand 0 predicate to
9758 nonimmediate_operand.
9759 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
9760 and corresponding splitters.
9761 (*fistdi2_1): Remove.
9763 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
9764 (lrintxfdi2): Remove expander. Reimplement as define_insn.
9765 (*fist<mode>2_1): Remove.
9766 (fist<mode>2): Ditto.
9767 (fist<mode>2_with_temp): Remove insn pattern and corresponding
9769 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
9770 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
9771 (fistdi2_<rounding>): Change operand 0 predicate to
9772 nonimmediate_operand.
9773 (fistdi2_<rounding>_with_temp): Remove insn pattern
9774 and corresponding splitters.
9775 (fist<mode>2_<rounding>): Change operand 0 predicate to
9776 nonimmediate_operand.
9777 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
9778 and corresponding splitters.
9780 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
9782 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9784 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
9787 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
9789 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9790 early gimple folding of vec_splat().
9791 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
9792 * gimple-fold.h: Add an extern define for tree_vec_extract().
9794 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
9796 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
9797 wrappers around TREE_TYPE comparisons.
9799 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
9802 * config/s390/predicates.md: Add nonsym_memory_operand.
9803 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
9804 contains a SYMBOL_REF, load it into an intermediate pseudo.
9805 (s390_emit_compare_and_swap): Legitimize operand.
9806 * config/s390/s390.md: Use the new nonsym_memory_operand
9807 with UNSPECV_CAS patterns.
9809 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
9812 * config/s390/s390-passes.def: New file.
9813 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
9815 (make_pass_s390_early_mach): Add declaration.
9816 * config/s390/s390.c (make_pass_s390_early_mach):
9817 (s390_option_override): Remove dynamic registration.
9818 * config/s390/t-s390: Add s390-passes.def.
9820 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
9822 * config/s390/s390.c (s390_decompose_constant_pool_ref):
9823 Remove UNSPEC_LTREL_BASE check.
9824 (annotate_constant_pool_refs): Likewise.
9825 (find_constant_pool_ref): Likewise.
9826 (find_ltrel_base): Removed.
9827 (replace_ltrel_base): Removed.
9828 (s390_mainpool_finish): Remove replace_ltrel_base call.
9829 (s390_chunkify_start): Remove pending LTREL_BASE logic.
9830 (s390_chunkify_finish): Remove replace_ltrel_base call.
9831 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
9833 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
9836 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
9837 to speculation_safe_value_not_needed.
9839 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
9840 Bernd Schmidt <bernds_cb1@t-online.de>
9842 * config/nvptx/nvptx-opts.h: New file.
9843 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
9844 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
9846 (TARGET_SM35): New macro.
9847 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
9849 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
9851 (misa=): New option.
9852 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
9854 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
9856 * config/i386/i386.md (truncdfsf2): Remove expander.
9857 (truncdfsf2_with_temp): Ditto.
9858 (truncxf<mode>2): Ditto.
9859 (*truncdfsf_fast_mixed): Remove insn pattern.
9860 (*truncdfsf_fast_i387): Ditto.
9861 (*truncdfsf_mixed): Ditto.
9862 (*truncdfsf_i387): Ditto.
9863 (*truncdfsf2_i387_1): Ditto.
9864 (*truncxfsf2_mixed): Ditto.
9865 (*truncxfdf2_mixed): Ditto.
9866 (*truncxf<mode>2_i387_noop): Ditto. Update callers
9867 to call gen_truncxf<mode>2 instead.
9868 (*truncxf<mode>2_i387): Remove.
9869 (reg->reg splitters): Remove splitter pattern.
9870 (reg->mem splitters): Ditto.
9872 (truncdfsf2): New insn pattern.
9873 (truncxf<mode>2): Ditto.
9875 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9877 * tree-ssa-math-opts.c (is_mult_by): New function.
9878 (is_square_of): Use the above.
9879 (optimize_recip_sqrt): New function.
9880 (pass_cse_reciprocals::execute): Use the above.
9882 2018-09-05 Richard Biener <rguenther@suse.de>
9885 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
9886 to zero-init the emplaced vec.
9888 2018-09-05 Martin Liska <mliska@suse.cz>
9890 PR tree-optimization/87205
9891 * tree-switch-conversion.c (pass_lower_switch::execute):
9892 Group cases for switch statements.
9894 2018-09-05 Richard Biener <rguenther@suse.de>
9896 PR tree-optimization/87217
9897 * tree-ssa-sccvn.c (vuse_valueize): New.
9898 (vn_reference_lookup_pieces): Use it.
9899 (vn_reference_lookup): Likewise.
9901 2018-09-05 Nathan Sidwell <nathan@acm.org>
9904 * stor-layout.c (place_field): Scan forwards to check last
9905 bitfield when ms_bitfield_placement is in effect.
9907 2018-09-05 Richard Biener <rguenther@suse.de>
9910 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
9913 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
9914 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
9916 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
9917 * config.gcc (extra_objs): Build it.
9918 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
9920 * config/aarch64/aarch64-passes.def
9921 (pass_tag_collision_avoidance): New pass.
9922 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
9923 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
9924 (aarch64_classify_address): Remove static qualifier.
9925 (aarch64_address_info, aarch64_address_type): Move to...
9926 * config/aarch64/aarch64-protos.h: ... here.
9927 (make_pass_tag_collision_avoidance): New function.
9928 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
9931 2018-09-05 Martin Liska <mliska@suse.cz>
9933 * doc/gcov.texi: Update documentation of humar
9935 * gcov.c (format_count): Print one decimal place, it provides
9936 more fine number of situations like '1G' vs. '1.4G'.
9938 2018-09-05 Martin Liska <mliska@suse.cz>
9941 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
9942 * optc-gen.awk: Allow 'Var' for Deprecated options in order
9943 to generate a MASK value.
9945 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
9948 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
9949 if frame pointer isn't used.
9950 (compute_frame_pointer_to_fb_displacement): Likewise.
9952 2018-09-04 Jakub Jelinek <jakub@redhat.com>
9955 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
9956 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
9957 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
9958 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
9959 and OPTION_MASK_ISA_XSAVEC_UNSET.
9961 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
9963 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
9964 XOR operations in NAND case.
9966 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
9968 * wide-int-range.cc (wide_int_range_convert): New.
9969 * wide-int-range.h (wide_int_range_convert): New.
9970 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
9971 code into wide_int_range_convert.
9972 (extract_range_into_wide_ints): Do not munge anti range constants
9973 into the entire domain. Just return the range back.
9975 2018-09-04 Martin Liska <mliska@suse.cz>
9977 * genmatch.c (output_line_directive): Add new argument
9979 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
9981 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
9983 * doc/invoke.texi (Option Summary): Add whitespace.
9985 * doc/invoke.texi (Option Summary): Add -Waligned-new.
9987 2018-09-04 Richard Biener <rguenther@suse.de>
9989 PR tree-optimization/87211
9990 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
9991 backedge value we're supposed to treat as VARYING also number
9992 the PHI to VARYING in case it got a different value-number already.
9994 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
9996 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
9997 (extract_range_from_binary_expr_1): Do not call
9998 vrp_can_optimize_bit_op.
9999 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
10001 (wide_int_range_get_mask_and_bounds): New.
10002 (wide_int_range_optimize_bit_op): New.
10003 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
10004 (wide_int_range_bit_and): Same.
10005 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
10006 (wide_int_range_optimize_bit_op): New.
10007 (wide_int_range_get_mask_and_bounds): New.
10009 2018-09-04 Richard Biener <rguenther@suse.de>
10011 PR tree-optimization/87176
10012 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
10013 variable. When value-numbering a virtual PHI node make sure
10014 to not value-number to the backedge value.
10016 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
10018 * doc/extend.texi (Long Long, Hex Floats): Document support for
10019 long long and hex floats in more recent versions of ISO C++.
10021 2018-09-03 Richard Biener <rguenther@suse.de>
10023 PR tree-optimization/87177
10024 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
10027 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
10029 * bb-reorder.c (edge_order): Convert to C-qsort-style
10030 tri-state comparator.
10031 (reorder_basic_blocks_simple): Change std::stable_sort to
10034 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
10036 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
10037 tri-state comparator.
10038 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
10040 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
10042 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
10043 (mergesort): ... here as maximum count for using netsort.
10044 (gcc_qsort): Set nlim to 3 if stable sort is requested.
10045 (gcc_stablesort): New.
10046 * system.h (gcc_stablesort): Declare.
10048 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
10050 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
10051 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
10052 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
10054 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
10056 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
10057 lxsdx and stxsdx alternatives.
10058 (*mov<mode>_hardfloat64): Ditto.
10059 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
10061 2018-09-03 Richard Biener <rguenther@suse.de>
10063 PR tree-optimization/87200
10064 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
10067 2018-09-03 Martin Liska <mliska@suse.cz>
10069 PR tree-optimization/87201
10070 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
10071 Fix parenthesis in an expression.
10073 2018-09-03 Richard Biener <rguenther@suse.de>
10075 PR tree-optimization/87197
10076 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
10077 visited. CSE the VN_INFO hashtable lookup.
10079 PR tree-optimization/87169
10080 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
10081 iterating make sure there's no extra backedges from irreducible
10082 regions feeding the header. Mark the destination block
10085 2018-09-03 Martin Liska <mliska@suse.cz>
10088 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
10089 * common/common-targhooks.c (default_get_valid_option_values):
10091 * common/common-targhooks.h (default_get_valid_option_values):
10093 * common/config/i386/i386-common.c: Move processor_target_table
10095 (ix86_get_valid_option_values): New function.
10096 (TARGET_GET_VALID_OPTION_VALUES): New macro.
10097 * config/i386/i386.c (struct ptt): Move to i386-common.c.
10098 (PTA_*): Move all defined masks into i386-common.c.
10099 (ix86_function_specific_restore): Use new processor_cost_table.
10100 * config/i386/i386.h (struct ptt): Moved from i386.c.
10101 (struct pta): Likewise.
10102 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
10103 * doc/tm.texi.in: Likewise.
10104 * opt-suggestions.c (option_proposer::suggest_option):
10105 Pass prefix to build_option_suggestions.
10106 (option_proposer::get_completions): Likewise.
10107 (option_proposer::build_option_suggestions): Use the new target
10109 * opts.c (struct option_help_tuple): New struct.
10110 (print_filtered_help): Use the new target hook.
10112 2018-09-03 Martin Liska <mliska@suse.cz>
10114 PR middle-end/59521
10115 * predict.c (set_even_probabilities): Add likely_edges
10116 argument and handle cases where we have precisely one
10118 (combine_predictions_for_bb): Catch also likely_edges.
10119 (tree_predict_by_opcode): Handle gswitch statements.
10120 * tree-cfg.h (find_case_label_for_value): New declaration.
10121 (find_taken_edge_switch_expr): Likewise.
10122 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
10123 Find pivot in decision tree based on probabily, not by number of
10126 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
10128 * doc/standards.texi (Standards): Update Objective-C reference.
10130 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
10132 * doc/install.texi (Prerequisites): Update link for MPC.
10134 2018-09-01 Michael Matz <matz@suse.de>
10136 PR tree-optimization/87074
10137 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
10138 PHIs for outer-loop uses.
10140 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
10142 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
10143 * doc/invoke.texi (C Dialect Options): Ditto.
10145 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
10147 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
10149 2018-08-31 Richard Biener <rguenther@suse.de>
10151 PR tree-optimization/87168
10152 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
10153 (rpo_elim::eliminate_avail): When OP was not visited it must
10156 2018-08-31 David Malcolm <dmalcolm@redhat.com>
10158 * tree-vrp.c (copy_value_range): Convert param "from" from
10159 "value_range *" to "const value_range *".
10160 (range_is_null): Likewise for param "vr".
10161 (range_int_cst_p): Likewise.
10162 (range_int_cst_singleton_p): Likewise.
10163 (symbolic_range_p): Likewise.
10164 (value_ranges_intersect_p): Likewise for both params.
10165 (value_range_nonnegative_p): Likewise for param "vr".
10166 (value_range_constant_singleton): Likewise.
10167 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
10168 (extract_range_into_wide_ints): Likewise for param "vr".
10169 (extract_range_from_multiplicative_op): Likewise for params "vr0"
10171 (vrp_can_optimize_bit_op): Likewise.
10172 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
10174 (extract_range_from_unary_expr): Likewise.
10175 (debug_value_range): Likewise for param "vr".
10176 (value_range::dump): Add "const" qualifier.
10177 (vrp_prop::check_array_ref): Convert local "vr" from
10178 "value_range *" to "const value_range *".
10179 (vrp_prop::check_mem_ref): Likewise.
10180 (vrp_prop::visit_stmt): Likewise for local "old_vr".
10181 (vrp_intersect_ranges_1): Likewise for param "vr_1".
10182 (vrp_intersect_ranges): Likewise.
10183 (simplify_stmt_for_jump_threading): Likewise for local "vr".
10184 (vrp_prop::vrp_finalize): Likewise.
10185 * tree-vrp.h (value_range::dump): Add "const" qualifier.
10186 (vrp_intersect_ranges): Add "const" qualifier to params as above.
10187 (extract_range_from_unary_expr): Likewise.
10188 (value_range_constant_singleton): Likewise.
10189 (symbolic_range_p): Likewise.
10190 (copy_value_range): Likewise.
10191 (extract_range_from_binary_expr_1): Likewise.
10192 (range_int_cst_p): Likewise.
10193 (vrp_set_zero_nonzero_bits): Likewise.
10194 (range_int_cst_singleton_p): Likewise.
10196 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
10198 * config/aarch64/arm_neon.h (vabsd_s64): New.
10199 (vnegd_s64): Likewise.
10201 2018-08-31 Martin Jambor <mjambor@suse.cz>
10203 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
10205 2018-08-31 Martin Liska <mliska@suse.cz>
10207 * ipa-icf.c (sem_item::add_type): Use
10208 sem_item::m_type_hash_cache.
10209 * ipa-icf.h: Move the cache from sem_item_optimizer
10212 2018-08-31 Nathan Sidwell <nathan@acm.org>
10214 * doc/extend.texi (Backwards Compatibility): Remove implicit
10215 extern C leeway of () being (...).
10217 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10219 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
10221 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
10225 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
10227 2018-08-31 Jakub Jelinek <jakub@redhat.com>
10229 PR middle-end/87138
10230 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
10231 gen_int_mode. Formatting fixes.
10233 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
10235 * target.def (custom_function_descriptors): Improve documentation.
10236 * doc/tm.texi.in (Trampolines): Expand discussion of function
10237 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
10238 beginning of the section.
10239 * doc/tm.texi: Regenerated.
10241 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
10243 * cfg.h (class auto_edge_flag): Spell out the template-id of the
10244 base class in the initializer list. This is a workaround for
10245 building with older GCC.
10246 (class auto_bb_flag): Likewise.
10248 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
10250 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
10251 (altivec_vcmpequ<VI_char>_p): Remove star.
10252 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
10254 (expand_strncmp_vec_sequence): New function.
10255 (emit_final_str_compare_vec): New function.
10256 (expand_strn_compare): Add support for vector strncmp.
10257 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
10258 length specification to bytes.
10259 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
10260 (vcmpnezb_p): New pattern.
10261 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
10262 for option -mstring-compare-inline-limit.
10264 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
10266 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
10267 (PTA_SKYLAKE): Add PTA_AES.
10268 (PTA_GOLDMONT): Likewise.
10270 2018-08-29 Jan Hubicka <jh@suse.cz>
10273 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
10274 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
10276 2018-08-29 Jan Hubicka <jh@suse.cz>
10278 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
10280 (hash_tree): Do not visit TYPE_STUB_DECL.
10281 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
10282 stream TYPE_STUB_DECL.
10283 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
10284 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
10285 after free_lang_data.
10286 (type_in_anonymous_namespace_p): Likewise.
10288 2018-08-29 Jan Hubicka <jh@suse.cz>
10290 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
10291 comment that it has to be even number.
10292 (class sreal): Change m_sig type to int32_t.
10293 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
10294 int64_t for temporary calculations.
10295 (sreal_verify_basics): Drop one bit from minimum and maximum.
10297 2018-08-30 Richard Biener <rguenther@suse.de>
10299 PR tree-optimization/87147
10300 * tree-ssa-sccvn.c (SSA_VISITED): New function.
10301 (visit_phi): When the degenerate result is from the backedge and
10302 we didn't visit its definition yet drop to VARYING.
10303 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
10305 2018-08-29 Jan Hubicka <jh@suse.cz>
10307 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
10309 (hash_tree): Likewise.
10311 2018-08-29 Jan Hubicka <jh@suse.cz>
10313 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
10314 and TYPE_NEXT_REF_TO.
10316 2018-08-29 Jan Hubicka <jh@suse.cz>
10318 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
10319 comment that it has to be even number.
10320 (class sreal): Change m_sig type to int32_t.
10321 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
10322 int64_t for temporary calculations.
10323 (sreal_verify_basics): Drop one bit from minimum and maximum.
10325 2018-08-30 Tamar Christina <tamar.christina@arm.com>
10327 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
10329 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
10331 PR middle-end/86995
10332 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
10333 if to_add is negative.
10335 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
10337 PR middle-end/87053
10338 * builtins.c (c_strlen): Improve range checks.
10340 2018-08-29 Martin Sebor <msebor@redhat.com>
10341 Jeff Law <law@redhat.com>
10343 PR tree-optimization/86714
10344 PR tree-optimization/86711
10345 * builtins.c (c_strlen): Add arguments to call to string_constant.
10346 * expr.c (string_constant): Add argument. Detect missing nul
10347 terminator and outermost declaration it's missing in.
10348 * expr.h (string_constant): Add argument.
10349 * fold-const.c (read_from_constant_string): Add arguments to call to
10351 (c_getstr): Likewise.
10352 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
10353 to string_constant.
10354 * tree-ssa-strlen.c (get_stridx): Likewise.
10356 2018-08-29 Jan Hubicka <jh@suse.cz>
10358 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
10359 Do not stream DECL_VINDEX.
10360 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
10361 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
10362 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
10364 2018-08-29 Richard Biener <rguenther@suse.de>
10366 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
10367 virtual operands that are not default defs to honor region
10369 (rpo_vn_valueize): Remove ineffective code here.
10371 2018-08-29 Richard Biener <rguenther@suse.de>
10373 PR tree-optimization/87132
10374 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
10375 when skipping defs reachable over backedges.
10377 2018-08-29 Richard Biener <rguenther@suse.de>
10379 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
10380 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
10381 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
10382 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
10383 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
10384 (vn_reference_lookup_3): Remove use of const_parms.
10385 (free_rpo_vn): Do not free const_parms.
10386 (do_rpo_vn): Do not call init_const_parms.
10387 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
10388 SSA_NAME_POINTS_TO_READONLY_MEMORY.
10389 (call_may_clobber_ref_p_1): Likewise.
10391 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
10394 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
10396 (-ftree-scev-cprop): Document.
10398 2018-08-29 Jan Hubicka <jh@suse.cz>
10400 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
10402 (sreal constructor): Update.
10403 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
10404 sreal:operator/): Update.
10406 2018-08-29 Martin Liska <mliska@suse.cz>
10408 * tree-switch-conversion.c (switch_conversion::expand):
10409 Strenghten assumption about gswitch statements.
10411 2018-08-29 Richard Biener <rguenther@suse.de>
10413 PR tree-optimization/87117
10414 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
10415 re-value-number released SSA VDEFs.
10417 2018-08-29 Richard Biener <rguenther@suse.de>
10419 PR tree-optimization/87126
10420 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
10422 2018-08-28 Jim Wilson <jimw@sifive.com>
10424 * config/riscv/pic.md: Rewrite.
10425 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
10427 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
10428 (SOFTF, default_load, softload, softstore): New.
10430 2018-08-28 Jeff Law <law@redhat.com>
10432 * fold-const.c (fold_binary_loc): Remove recently added assert.
10434 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
10436 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
10437 to OP parmeter of generated function.
10439 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
10441 PR tree-optimization/87009
10442 * match.pd: Add boolean optimizations.
10444 2018-08-28 Martin Sebor <msebor@redhat.com>
10446 PR middle-end/86631
10447 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
10448 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
10449 (pass_walloca::gate): Use it.
10450 (alloca_call_type): Same.
10451 (pass_walloca::execute): Same.
10452 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
10454 2018-08-28 David Malcolm <dmalcolm@redhat.com>
10456 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
10457 GCC_VERSION for usage of "__gcc_dump_printf__" format from
10458 >= 3005 to >= 9000.
10460 2018-08-28 Richard Biener <rguenther@suse.de>
10462 PR tree-optimization/87124
10463 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
10464 constants before looking up avail.
10466 2018-08-28 Jakub Jelinek <jakub@redhat.com>
10468 PR middle-end/87099
10469 * calls.c (maybe_warn_nonstring_arg): Punt early if
10470 warn_stringop_overflow is zero. Don't call get_range_strlen
10471 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
10472 Swap comparison operands to have constants on rhs. Only use
10473 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
10474 increment lenrng[0].
10476 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
10478 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
10479 use of tree_to_shwi. Remove duplicated test for the size being
10480 a whole number of bytes.
10482 2018-08-28 Richard Biener <rguenther@suse.de>
10484 PR tree-optimization/87117
10485 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
10486 Handle removed stmt without LHS (GIMPLE_NOP).
10488 2018-08-28 Richard Biener <rguenther@suse.de>
10490 PR tree-optimization/87117
10491 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
10492 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
10494 2018-08-28 Richard Biener <rguenther@suse.de>
10496 PR tree-optimization/87117
10497 * tree-ssa-pre.c (compute_avail): Do not make expressions
10498 with predicated values available.
10499 (get_expr_value_id): Assert we do not run into predicated value
10502 2018-08-28 Richard Biener <rguenther@suse.de>
10504 PR tree-optimization/87117
10505 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
10506 get virtual operands.
10507 (get_expr_operands): Handle STRING_CST like other decls.
10509 2018-08-28 Martin Liska <mliska@suse.cz>
10511 * tree.h: Update documentation of fndecl_built_in_p
10515 2018-08-27 Jeff Law <law@redhat.com>
10516 PR tree-optimization/87110
10517 * tree-ssa-dse.c (compute_trims): Handle non-constant
10520 2018-08-27 Martin Sebor <msebor@redhat.com>
10522 PR tree-optimization/86914
10523 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
10525 2018-08-27 Martin Sebor <msebor@redhat.com>
10527 PR tree-optimization/87112
10528 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
10529 the type of the bound argument.
10531 2018-08-27 Jeff Law <law@redhat.com>
10533 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
10534 type does not have a TYPE_SIZE_UNIT.
10536 2018-08-27 Steve Ellcey <sellcey@cavium.com>
10538 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
10539 with include of backend.h.
10541 2018-08-27 Richard Biener <rguenther@suse.de>
10543 PR tree-optimization/86927
10544 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
10545 use const cond reduction code.
10547 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
10549 PR tree-optimization/85758
10550 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
10552 2018-08-27 David Malcolm <dmalcolm@redhat.com>
10555 * diagnostic-show-locus.c (class layout_range): Update for
10556 conversion of show_caret_p to a tri-state.
10557 (layout_range::layout_range): Likewise.
10558 (make_range): Likewise.
10559 (layout::maybe_add_location_range): Likewise.
10560 (layout::should_print_annotation_line_p): Don't show annotation
10561 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
10562 (layout::get_state_at_point): Update for conversion of
10563 show_caret_p to a tri-state. Bail out early for
10564 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
10565 underlining or source colorization.
10566 (gcc_rich_location::add_location_if_nearby): Update for conversion
10567 of show_caret_p to a tri-state.
10568 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
10569 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
10571 (selftest::test_one_liner_labels): Likewise.
10572 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
10573 conversion of show_caret_p to a tri-state.
10574 * pretty-print.c (text_info::set_location): Likewise.
10575 * pretty-print.h (text_info::set_location): Likewise.
10576 * substring-locations.c (format_warning_n_va): Likewise.
10577 * tree-diagnostic.c (default_tree_printer): Likewise.
10578 * tree-pretty-print.c (newline_and_indent): Likewise.
10580 2018-08-27 David Malcolm <dmalcolm@redhat.com>
10583 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
10584 line above for line-insertion fix-it hints.
10585 (selftest::test_fixit_insert_containing_newline): Update the
10586 expected results, and add a test with line-numbering enabled.
10588 2018-08-27 Martin Liska <mliska@suse.cz>
10591 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
10592 params with DECL_HAS_VALUE_EXPR_P.
10594 2018-08-27 Martin Liska <mliska@suse.cz>
10596 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
10597 selected expansion strategy.
10599 2018-08-27 Martin Liska <mliska@suse.cz>
10601 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
10602 * builtins.c (is_builtin_fn): Likewise.
10603 * attribs.c (diag_attr_exclusions): Use new function
10604 fndecl_built_in_p and remove check for FUNCTION_DECL if
10606 (builtin_mathfn_code): Likewise.
10607 (fold_builtin_expect): Likewise.
10608 (fold_call_expr): Likewise.
10609 (fold_builtin_call_array): Likewise.
10610 (fold_call_stmt): Likewise.
10611 (set_builtin_user_assembler_name): Likewise.
10612 (is_simple_builtin): Likewise.
10613 * calls.c (gimple_alloca_call_p): Likewise.
10614 (maybe_warn_nonstring_arg): Likewise.
10615 * cfgexpand.c (expand_call_stmt): Likewise.
10616 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
10617 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
10618 (cgraph_node::verify_node): Likewise.
10619 * cgraphclones.c (build_function_decl_skip_args): Likewise.
10620 (cgraph_node::create_clone): Likewise.
10621 * config/arm/arm.c (arm_insert_attributes): Likewise.
10622 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
10623 * dse.c (scan_insn): Likewise.
10624 * expr.c (expand_expr_real_1): Likewise.
10625 * fold-const.c (operand_equal_p): Likewise.
10626 (fold_binary_loc): Likewise.
10627 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
10628 * gimple-low.c (lower_stmt): Likewise.
10629 * gimple-pretty-print.c (dump_gimple_call): Likewise.
10630 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
10632 * gimple.c (gimple_build_call_from_tree): Likewise.
10633 (gimple_call_builtin_p): Likewise.
10634 (gimple_call_combined_fn): Likewise.
10635 * gimplify.c (gimplify_call_expr): Likewise.
10636 (gimple_boolify): Likewise.
10637 (gimplify_modify_expr): Likewise.
10638 (gimplify_addr_expr): Likewise.
10639 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
10640 * ipa-cp.c (determine_versionability): Likewise.
10641 * ipa-fnsummary.c (compute_fn_summary): Likewise.
10642 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
10643 * ipa-split.c (visit_bb): Likewise.
10644 (split_function): Likewise.
10645 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
10646 * lto-cgraph.c (input_node): Likewise.
10647 * lto-streamer-out.c (write_symbol): Likewise.
10648 * omp-low.c (setjmp_or_longjmp_p): Likewise.
10649 (lower_omp_1): Likewise.
10650 * predict.c (strip_predict_hints): Likewise.
10651 * print-tree.c (print_node): Likewise.
10652 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
10653 * trans-mem.c (is_tm_irrevocable): Likewise.
10654 (is_tm_load): Likewise.
10655 (is_tm_simple_load): Likewise.
10656 (is_tm_store): Likewise.
10657 (is_tm_simple_store): Likewise.
10658 (is_tm_abort): Likewise.
10659 (tm_region_init_1): Likewise.
10660 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
10661 * tree-cfg.c (verify_gimple_call): Likewise.
10662 (move_stmt_r): Likewise.
10663 (stmt_can_terminate_bb_p): Likewise.
10664 * tree-eh.c (lower_eh_constructs_2): Likewise.
10665 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
10666 * tree-inline.c (remap_gimple_stmt): Likewise.
10667 (copy_bb): Likewise.
10668 (estimate_num_insns): Likewise.
10669 (fold_marked_statements): Likewise.
10670 * tree-sra.c (scan_function): Likewise.
10671 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
10672 (optimize_stack_restore): Likewise.
10673 (pass_fold_builtins::execute): Likewise.
10674 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10675 (mark_all_reaching_defs_necessary_1): Likewise.
10676 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
10677 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
10678 (pass_forwprop::execute): Likewise.
10679 * tree-ssa-loop-im.c (stmt_cost): Likewise.
10680 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
10681 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
10682 * tree-ssa-strlen.c (get_string_length): Likewise.
10683 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
10684 (find_func_aliases_for_call): Likewise.
10685 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
10686 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
10687 * tree-tailcall.c (find_tail_calls): Likewise.
10688 * tree.c (need_assembler_name_p): Likewise.
10689 (free_lang_data_in_decl): Likewise.
10690 (get_call_combined_fn): Likewise.
10691 * ubsan.c (is_ubsan_builtin_p): Likewise.
10692 * varasm.c (incorporeal_function_p): Likewise.
10693 * tree.h (DECL_BUILT_IN): Remove and replace with
10695 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
10696 (fndecl_built_in_p): New.
10698 2018-08-27 Martin Liska <mliska@suse.cz>
10700 PR tree-optimization/86847
10701 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
10702 Dump also subtree probability.
10703 (switch_decision_tree::do_jump_if_equal): New function.
10704 (switch_decision_tree::emit_case_nodes): Handle special
10705 situations in balanced tree that can be emitted much simpler.
10706 Fix calculation of probabilities that happen in tree expansion.
10707 * tree-switch-conversion.h (struct cluster): Add
10709 (struct simple_cluster): Likewise.
10710 (struct case_tree_node): Add new function has_child.
10711 (do_jump_if_equal): New.
10713 2018-08-27 Martin Liska <mliska@suse.cz>
10715 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
10716 Add new argument to bit_test_cluster constructor.
10717 (bit_test_cluster::emit): Set bits really number of values
10719 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
10720 probability argument.
10721 * tree-switch-conversion.h (struct bit_test_cluster):
10722 Add m_handles_entire_switch.
10724 2018-08-27 Martin Liska <mliska@suse.cz>
10726 PR tree-optimization/86702
10727 * tree-switch-conversion.c (jump_table_cluster::emit):
10728 Make probabilities even for values in jump table
10729 according to number of cases handled.
10730 (switch_decision_tree::compute_cases_per_edge): Pass
10731 argument to reset_out_edges_aux function.
10732 (switch_decision_tree::analyze_switch_statement): Likewise.
10733 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
10736 2018-08-27 Martin Liska <mliska@suse.cz>
10738 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
10739 cfun argument explicitly.
10740 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
10741 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
10742 function gimple_switch_default_bb.
10743 (convert_switch_statements):
10745 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
10746 * stmt.c (label_to_block_fn): Use label_to_block and pass
10747 cfun argument explicitly and use gimple_switch_label_bb.
10748 (expand_case): Likewise.
10749 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
10750 cfun argument explicitly. Likewise.
10751 (make_edges_bb): Likewise.
10752 (make_cond_expr_edges): Likewise.
10753 (get_cases_for_edge): Likewise.
10754 (make_gimple_switch_edges): Likewise.
10755 (label_to_block_fn): Likewise.
10756 (label_to_block): Likewise.
10757 (make_goto_expr_edges): Likewise.
10758 (make_gimple_asm_edges): Likewise.
10759 (main_block_label): Likewise.
10760 (group_case_labels_stmt): Likewise.
10761 (find_taken_edge_computed_goto): Likewise.
10762 (find_taken_edge_switch_expr): Likewise.
10763 (gimple_verify_flow_info): Likewise.
10764 (gimple_redirect_edge_and_branch): Likewise.
10765 (gimple_switch_label_bb): New function.
10766 (gimple_switch_default_bb): Likewise.
10767 (gimple_switch_edge): Likewise.
10768 (gimple_switch_default_edge): Likewise.
10769 * tree-cfg.h (label_to_block_fn): Remove and replace ...
10770 (label_to_block): ... with this.
10771 (gimple_switch_label_bb): New.
10772 (gimple_switch_default_bb): Likewise.
10773 (gimple_switch_edge): Likewise.
10774 (gimple_switch_default_edge): Likewise.
10775 * tree-cfgcleanup.c (convert_single_case_switch): Use
10776 new gimple functions and pass new argument to label_to_block.
10777 (cleanup_control_flow_bb):
10778 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
10779 cfun argument explicitly.
10780 (make_eh_edges): Likewise.
10781 (redirect_eh_dispatch_edge): Likewise.
10782 (lower_resx): Likewise.
10783 (lower_eh_dispatch): Likewise.
10784 (maybe_remove_unreachable_handlers): Likewise.
10785 (unsplit_eh): Likewise.
10786 (cleanup_empty_eh): Likewise.
10787 (verify_eh_edges): Likewise.
10788 (verify_eh_dispatch_edge): Likewise.
10789 * tree-ssa-dom.c (record_edge_info): Likewise.
10790 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
10791 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
10792 (thread_through_normal_block): Likewise.
10793 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
10794 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
10795 * tree-switch-conversion.c (switch_conversion::collect): Use new
10797 (switch_conversion::check_final_bb): Likewise.
10798 (switch_conversion::gather_default_values): Pass new argument
10800 (switch_conversion::build_constructors): Likewise.
10801 (switch_decision_tree::compute_cases_per_edge): Use new
10802 gimple_switch_edge function.
10803 (switch_decision_tree::analyze_switch_statement): Pass new argument
10805 (switch_decision_tree::try_switch_expansion): Use
10806 gimple_switch_default_edge.
10807 * tree-vrp.c (find_switch_asserts): Pass new argument
10809 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
10810 (vr_values::simplify_switch_using_ranges): Likewise.
10812 2018-08-27 Richard Biener <rguenther@suse.de>
10814 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
10815 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
10817 * tree-ssa-sccvn.h (struct vn_pval): New structure.
10818 (struct vn_nary_op_s): Add unwind_to member. Add
10819 predicated_values flag and put result into a union together
10820 with a linked list of vn_pval.
10821 (struct vn_ssa_aux): Add name member to make maintaining
10822 a map of SSA name to vn_ssa_aux possible. Remove no longer
10823 needed info, dfsnum, low, visited, on_sccstack, use_processed
10824 and range_info_anti_range_p members.
10825 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
10826 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
10828 (vn_valueize): New global.
10829 (vn_context_bb): Likewise.
10830 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
10831 VN_INFO_PTR_INFO): Remove.
10832 * tree-ssa-sccvn.c: ... (rewrite)
10833 (pass_fre::execute): For -O2+ initialize loops and run
10834 RPO VN in optimistic mode (iterating). For -O1 and -Og
10835 run RPO VN in non-optimistic mode.
10836 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
10837 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
10838 * doc/invoke.texi (sccvn-max-scc-size): Remove.
10839 (rpo-vn-max-loop-depth): Document.
10840 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
10841 when valuezing the VUSE signals we walked out of the region.
10842 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
10843 (phi_translate): Set VN context block to use for availability
10845 (compute_avail): Likewise.
10846 (pre_valueize): New function.
10847 (pass_pre::execute): Adjust to the RPO VN API.
10849 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
10850 (propagate_constants_for_unrolling): Remove.
10851 (tree_unroll_loops_completely): Perform value-numbering
10852 on the unrolled bodies loop parent.
10854 2018-08-27 Richard Biener <rguenther@suse.de>
10856 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
10857 for partial antic compute.
10859 2018-08-27 Jakub Jelinek <jakub@redhat.com>
10861 PR rtl-optimization/87065
10862 * combine.c (simplify_if_then_else): Formatting fix.
10863 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
10865 (known_cond): Don't return const_true_rtx for vector modes. Use
10866 CONST0_RTX instead of const0_rtx. Formatting fixes.
10868 2018-08-27 Martin Liska <mliska@suse.cz>
10870 PR gcov-profile/87069
10871 * gcov.c (process_file): Record files already processed
10872 and warn about a file being processed multiple times.
10874 2018-08-27 Martin Liska <mliska@suse.cz>
10877 * config/aarch64/aarch64.c (aarch64_override_options_internal):
10878 Set default values for x_aarch64_*_string strings.
10879 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
10880 prefix. For -mabi do not print '=ABI' in help and use
10881 <option_value> format for -msve-vector-bits and -moverride
10884 2018-08-26 Jeff Law <law@redhat.com>
10886 * config/mips/frame-header-opt.c: Include "backend.h" rather than
10889 2018-08-26 Marek Polacek <polacek@redhat.com>
10891 PR c++/87029, Implement -Wredundant-move.
10892 * doc/invoke.texi: Document -Wredundant-move.
10894 2018-08-25 Martin Sebor <msebor@redhat.com>
10896 PR tree-optimization/87059
10897 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
10898 to the same type as the other.
10899 * fold-const.c (fold_binary_loc): Assert expectation.
10901 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
10903 * config/darwin.c (machopic_legitimize_pic_address): Clean up
10904 extraneous parentheses, dead code section and formatting.
10906 2018-08-24 David Malcolm <dmalcolm@redhat.com>
10909 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
10910 wide enough for jumps in the line-numbering to be visible.
10911 (layout::print_gap_in_line_numbering): New member function.
10912 (layout::calculate_line_spans): When using line numbering, merge
10913 line spans that are only 1 line apart.
10914 (diagnostic_show_locus): When printing line numbers, show gaps in
10915 line numbering directly, rather than printing headers.
10916 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
10917 line-numbering with multiple line spans.
10918 (selftest::test_fixit_insert_containing_newline_2): Add test of
10919 line-numbering, in which the spans are close enough to be merged.
10921 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
10923 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
10924 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
10925 * tree-vrp.c (range_is_nonnull): Remove.
10926 (range_includes_zero_p): Accept value_range instead of min/max.
10927 (extract_range_from_binary_expr_1): Do not early bail on
10929 Use range_includes_zero_p instead of range_is_nonnull.
10930 (extract_range_from_unary_expr): Use range_includes_zero_p instead
10931 of range_is_nonnull.
10932 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
10933 special case VR_ANTI_RANGE.
10934 (vrp_finalize): Same.
10935 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
10936 instead of min/max.
10937 (range_is_nonnull): Remove.
10938 * vr-values.c (vrp_stmt_computes_nonzero): Use
10939 range_includes_zero_p instead of range_is_nonnull.
10940 (extract_range_basic): Pass value_range to range_includes_zero_p
10941 instead of range_is_nonnull.
10943 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
10945 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
10946 * emit-rtl.h (rtl_data): Remove return_bnd.
10947 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
10948 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
10949 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
10950 (POINTER_BOUNDS_MODE): Remove definition.
10951 (make_pointer_bounds_mode): Remove.
10952 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
10953 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
10954 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
10955 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
10956 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
10957 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
10958 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
10960 * config/i386/i386-modes.def (BND32, BND64): Remove.
10961 * config/i386/i386.c (dbx_register_map): Remove bound registers.
10962 (dbx64_register_map): Ditto.
10963 (svr4_dbx_register_map): Ditto.
10964 (indirect_thunk_bnd_needed): Remove.
10965 (indirect_thunks_bnd_used): Ditto.
10966 (indirect_return_bnd_needed): Ditto.
10967 (indirect_return_via_cx_bnd): Ditto.
10968 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
10969 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
10970 (output_indirect_thunk): Ditto. Remove need_prefix argument.
10971 (output_indirect_thunk_function): Remove handling of
10972 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
10973 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
10974 (ix86_save_reg): Remove handling of crtl->return_bnd.
10975 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
10976 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
10977 and UNSPEC_BNDLX_ADDR.
10978 (ix86_output_indirect_branch_via_reg): Remove handling of
10979 indirect_thunk_prefix_bnd.
10980 (ix86_output_indirect_branch_via_push): Ditto.
10981 (ix86_output_function_return): Ditto.
10982 (ix86_output_indirect_function_return): Ditto.
10983 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
10984 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
10985 (CALL_USED_REGISTERS): Ditto.
10986 (REG_ALLOC_ORDER): Update for removal of bound registers.
10987 (HI_REGISTER_NAMES): Ditto.
10988 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
10989 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
10990 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
10991 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
10992 (FIRST_PSEUDO_REG): Update.
10993 (BND): Remove mode iterator.
10994 * config/i386/predicates.md (bnd_mem_operator): Remove.
10996 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
10998 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
11001 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
11003 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
11004 the case in which the permute needs only a single element and
11005 repeats for every vector of the result. Extend that case to
11006 handle variable-length vectors.
11007 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
11009 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
11012 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
11013 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
11015 2018-08-24 Richard Biener <rguenther@suse.de>
11017 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
11018 bb_flags_allocated members.
11019 (auto_flag): New RAII class for allocating flags.
11020 (auto_edge_flag): New RAII class for allocating edge flags.
11021 (auto_bb_flag): New RAII class for allocating bb flags.
11022 * cfgloop.c (verify_loop_structure): Allocate temporary edge
11024 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
11025 in favor of temporarily allocated BB flag.
11026 * hsa-brig.c: Re-order includes.
11027 * hsa-dump.c: Likewise.
11028 * hsa-regalloc.c: Likewise.
11029 * print-rtl.c: Likewise.
11030 * profile-count.c: Likewise.
11032 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
11035 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
11038 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
11041 * wide-int-range.cc (wide_int_range_div): Do not ignore result
11042 from wide_int_range_multiplicative_op.
11044 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11046 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
11049 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
11051 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
11054 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
11056 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
11057 full GENERIC expression used for replacement.
11059 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
11061 * tree-vrp.c (abs_extent_range): Remove.
11062 (extract_range_into_wide_ints): Pass wide ints by reference.
11063 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
11064 Pass wide ints by reference in all calls to
11065 extract_range_into_wide_ints.
11066 * wide-int-range.cc (wide_int_range_div): New.
11067 * wide-int-range.h (wide_int_range_div): New.
11068 (wide_int_range_includes_zero_p): New.
11069 (wide_int_range_zero_p): New.
11071 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
11073 * config/aarch64/aarch64.md (arches): New enum.
11074 (arch): New enum attr.
11075 (arch_enabled): New attr.
11076 (enabled): Now uses arch_enabled only.
11077 (simd, sve, fp16): Removed attribute.
11078 (fp): Attr now defined in terms of 'arch'.
11079 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
11080 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
11081 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
11082 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
11083 attributes into 'arch'.
11084 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
11085 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
11086 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
11087 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
11088 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
11089 'simd' attribute into 'arch'.
11090 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
11091 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
11092 Convert use of 'fp' attribute to 'arch'.
11093 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
11094 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
11096 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
11097 (different modes) Merge 'fp' and 'simd' into 'arch'.
11098 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
11099 'simd' into 'arch'.
11101 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
11103 PR rtl-optimization/87026
11104 * expmed.c (canonicalize_comparison): If we can no longer create
11105 pseudoregisters, don't.
11107 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
11110 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
11112 * config/arm/arm.c (speculation_barrier_libfunc): New static
11114 (arm_init_libfuncs): Initialize it.
11115 (arm_emit_speculation_barrier): New function.
11116 * config/arm/arm.md (speculation_barrier): Call
11117 arm_emit_speculation_barrier for architectures that do not have
11119 (speculation_barrier_insn): Only match on Armv7 or later.
11121 2018-08-23 Richard Biener <rguenther@suse.de>
11123 PR middle-end/87024
11124 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
11127 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
11129 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
11130 of single-vector TBLs.
11131 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
11132 one input is given.
11134 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
11137 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
11138 aarch64_evpc_tbl guard.
11140 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
11142 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
11145 2018-08-22 Martin Sebor <msebor@redhat.com>
11147 PR middle-end/87052
11148 * tree-pretty-print.c (pretty_print_string): Add argument.
11149 (dump_generic_node): Call to pretty_print_string with string size.
11151 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
11153 PR rtl-optimization/86771
11154 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
11155 of two SETs into those two SETs, one to be placed at i2, if that SETs
11156 destination is modified between i2 and i3.
11158 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
11160 PR tree-optimization/86725
11161 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
11163 (vect_analyze_scalar_cycles_1): Check it.
11165 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
11167 PR tree-optimization/86725
11168 * tree-vect-loop.c (vect_is_simple_reduction): When treating
11169 an outer loop phi as a double reduction, make sure that the
11170 single user of the phi result is an inner loop phi.
11172 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
11174 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
11175 grouped stores with gaps to a strided group.
11177 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
11179 * tree-vect-stmts.c (get_group_load_store_type)
11180 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
11181 first statement in a group.
11183 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
11185 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
11186 the sequence used in gcc/gcc.c.
11188 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
11191 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
11194 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
11196 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
11197 Darwin10-specific unwinder-shim.
11198 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
11199 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
11200 New to cater for Darwin10 Rosetta.
11202 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
11204 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
11205 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
11207 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
11212 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
11213 (dwarf2out_switch_text_section): Generate a local label for the
11214 second function sub-section and apply it as the second FDE start
11216 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
11217 second sub-section start.
11219 2018-08-22 Richard Biener <rguenther@suse.de>
11221 PR tree-optimization/86988
11222 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
11224 2018-08-22 Richard Biener <rguenther@suse.de>
11226 PR tree-optimization/86945
11227 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
11229 2018-08-22 Alexandre Oliva <oliva@adacore.com>
11231 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
11232 a comment about how uses of r2 for .sdata2 come about.
11234 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
11236 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
11238 2018-08-21 Marek Polacek <polacek@redhat.com>
11240 PR c++/86981, Implement -Wpessimizing-move.
11241 * doc/invoke.texi: Document -Wpessimizing-move.
11243 2018-08-21 Jan Hubicka <jh@suse.cz>
11245 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
11246 * tree.h (is_redundant_typedef): Remove.
11247 * dwarf2out.c (is_redundant_typedef): Turn into static function.
11249 2018-08-21 Jan Hubicka <jh@suse.cz>
11251 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
11254 2018-08-21 Tamar Christina <tamar.christina@arm.com>
11256 * expmed.c (extract_low_bits): Reject invalid subregs early.
11258 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
11260 PR middle-end/86121
11261 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
11264 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
11266 * config/vxworks.h: Guard vxworks_asm_out_constructor and
11267 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
11268 * config/vxworks.c: Likewise.
11270 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
11272 * config/vxworks.c: Set targetm.have_ctors_dtors
11273 if HAVE_INITFINI_ARRAY_SUPPORT.
11274 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
11275 if HAVE_INITFINI_ARRAY_SUPPORT.
11277 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
11279 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
11280 default search path for VxWorks < 7.
11282 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
11284 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
11285 (get_format_string): Refer to c_getstr.
11287 2018-08-21 Tom de Vries <tdevries@suse.de>
11289 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
11290 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
11291 (debuginfo_early_stop): Declare.
11292 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
11293 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
11294 (debuginfo_early_stop): New function.
11295 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
11296 and debuginfo_early_stop.
11297 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
11298 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
11299 (general_init): Call debuginfo_early_init.
11300 (finalize): Call debuginfo_fini.
11301 (do_compile): Call debuginfo_init.
11302 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
11303 -fdump-early-debug.
11304 (@item -fdump-debug, @item -fdump-earlydebug): Add.
11306 2018-08-21 Tom de Vries <tdevries@suse.de>
11308 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
11309 flag_dump_noaddr and flag_dump_unnumbered.
11311 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
11313 * wide-int-range.cc (wide_int_range_abs): New.
11314 (wide_int_range_order_set): Rename from wide_int_range_min_max.
11315 * wide-int-range.h (wide_int_range_abs): New.
11316 (wide_int_range_min_max): New.
11317 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
11318 case to call wide_int_range_abs.
11319 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
11320 (extract_range_from_abs_expr): Delete.
11322 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
11325 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
11326 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
11329 2018-08-20 Nathan Sidwell <nathan@acm.org>
11330 Jeff Law <law@redhat.com>
11332 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
11333 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
11335 2018-08-20 David Malcolm <dmalcolm@redhat.com>
11338 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
11339 (decl_attributes): Likewise.
11340 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
11342 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
11343 * diagnostic-core.h (class auto_diagnostic_group): New class.
11344 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
11345 (diagnostic_report_diagnostic): Handle the first diagnostics within
11347 (emit_diagnostic): Add auto_diagnostic_group instance.
11348 (inform): Likewise.
11349 (inform_n): Likewise.
11350 (warning): Likewise.
11351 (warning_at): Likewise.
11352 (warning_n): Likewise.
11353 (pedwarn): Likewise.
11354 (permerror): Likewise.
11356 (error_n): Likewise.
11357 (error_at): Likewise.
11359 (fatal_error): Likewise.
11360 (internal_error): Likewise.
11361 (internal_error_no_backtrace): Likewise.
11362 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
11363 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
11364 * diagnostic.h (struct diagnostic_context): Add fields
11365 "diagnostic_group_nesting_depth",
11366 "diagnostic_group_emission_count", "begin_group_cb",
11368 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
11369 Add auto_diagnostic_group instance(s).
11370 (find_explicit_erroneous_behavior): Likewise.
11371 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
11372 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
11373 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
11374 (gimplify_va_arg_expr): Likewise.
11375 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
11376 (HSA_SORRY_AT): Likewise.
11377 * ipa-devirt.c (compare_virtual_tables): Likewise.
11378 (warn_odr): Likewise.
11379 * multiple_target.c (expand_target_clones): Likewise.
11380 * opts-common.c (cmdline_handle_error): Likewise.
11381 * reginfo.c (globalize_reg): Likewise.
11382 * substring-locations.c (format_warning_n_va): Likewise.
11383 * tree-inline.c (expand_call_inline): Likewise.
11384 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
11385 * tree-ssa-loop-niter.c
11386 (do_warn_aggressive_loop_optimizations): Likewise.
11387 * tree-ssa-uninit.c (warn_uninit): Likewise.
11388 * tree.c (warn_deprecated_use): Likewise.
11390 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
11393 * config/i386/i386.md (eh_return): Always update EH return
11394 address in word_mode.
11396 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
11398 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
11399 TARGET_SPLIT_COMPLEX_ARG is defined.
11401 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11403 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
11405 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11408 * expr.c (expand_assignment): Assert that bitpos is positive.
11409 (store_field): Likewise
11410 (expand_expr_real_1): Make sure that bitpos is positive.
11411 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
11414 2018-08-20 Nathan Sidwell <nathan@acm.org>
11416 * Makefile.in (CPP_ID_DATA_H): Delete.
11417 (CPP_INTERNAL_H): Don't add it.
11418 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
11419 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
11421 2018-08-20 Richard Biener <rguenther@suse.de>
11423 PR tree-optimization/78655
11424 * tree-vrp.c (extract_range_from_binary_expr_1): Make
11425 pointer + offset nonnull if either operand is nonnull work.
11427 2018-08-20 Tom de Vries <tdevries@suse.de>
11429 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
11430 unless the referenced die describes the added property using
11431 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
11432 Otherwise, add a DW_AT_location to the referenced die.
11434 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
11437 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
11438 register_operand when calling ix86_set_reg_reg_cost.
11439 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
11440 Set *total to 0 for operands that satisfy x86_64_immediate_operand
11441 predicate and to 1 otherwise.
11443 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
11445 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
11446 emit a diagnostic that it is not supported and reset the option.
11447 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
11448 supported and consume the option. (ASM_FINAL_SPEC): New.
11450 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
11452 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
11455 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
11457 C-SKY port: Documentation
11459 * doc/extend.texi (C-SKY Function Attributes): New section.
11460 * doc/invoke.texi (Option Summary): Add C-SKY options.
11461 (C-SKY Options): New section.
11462 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
11464 2018-08-17 Jojo <jijie_rong@c-sky.com>
11465 Huibin Wang <huibin_wang@c-sky.com>
11466 Sandra Loosemore <sandra@codesourcery.com>
11467 Chung-Lin Tang <cltang@codesourcery.com>
11469 C-SKY port: Backend implementation
11471 * config/csky/*: New.
11472 * common/config/csky/*: New.
11474 2018-08-17 Jojo <jijie_rong@c-sky.com>
11475 Huibin Wang <huibin_wang@c-sky.com>
11476 Sandra Loosemore <sandra@codesourcery.com>
11477 Chung-Lin Tang <cltang@codesourcery.com>
11478 Andrew Jenner <andrew@codesourcery.com>
11480 C-SKY port: Configury
11482 * config.gcc (csky-*-*): New.
11483 * configure.ac: Add csky to targets for dwarf2 debug_line support.
11484 * configure: Regenerated.
11486 2018-08-17 David Malcolm <dmalcolm@redhat.com>
11488 * dump-context.h: Include "dumpfile.h".
11489 (dump_context::dump_printf_va): Convert final param from va_list
11490 to va_list *. Convert from ATTRIBUTE_PRINTF to
11491 ATTRIBUTE_GCC_DUMP_PRINTF.
11492 (dump_context::dump_printf_loc_va): Likewise.
11493 * dumpfile.c: Include "stringpool.h".
11494 (make_item_for_dump_printf_va): Delete.
11495 (make_item_for_dump_printf): Delete.
11496 (class dump_pretty_printer): New class.
11497 (dump_pretty_printer::dump_pretty_printer): New ctor.
11498 (dump_pretty_printer::emit_items): New member function.
11499 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
11501 (dump_pretty_printer::emit_item): New member function.
11502 (dump_pretty_printer::stash_item): New member function.
11503 (dump_pretty_printer::format_decoder_cb): New member function.
11504 (dump_pretty_printer::decode_format): New member function.
11505 (dump_context::dump_printf_va): Reimplement in terms of
11506 dump_pretty_printer.
11507 (dump_context::dump_printf_loc_va): Convert final param from va_list
11509 (dump_context::begin_scope): Reimplement call to
11510 make_item_for_dump_printf.
11511 (dump_printf): Update for change to dump_printf_va.
11512 (dump_printf_loc): Likewise.
11513 (selftest::test_capture_of_dump_calls): Convert "stmt" from
11514 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
11515 with %T, %E, and %G.
11516 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
11517 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
11518 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
11519 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
11520 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
11521 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
11522 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
11523 within a dump_printf_loc call to "%wu".
11524 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
11525 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
11526 missing space after "=".
11527 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
11528 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
11529 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
11530 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
11531 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
11532 duplicate "vectorized" from message.
11534 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
11536 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
11537 polyNxK_t element's TYPE_STRING_FLAG.
11539 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
11541 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
11542 (they were unnamed before). Fix comments.
11544 2018-08-17 Nathan Sidwell <nathan@acm.org>
11546 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
11548 2018-08-17 Richard Biener <rguenther@suse.de>
11550 PR tree-optimization/86841
11551 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
11553 2018-08-17 Martin Liska <mliska@suse.cz>
11555 * common.opt: Remove Warn, Init and Report for options with
11556 Ignore/Deprecated flag. Warning is done automatically for
11558 * config/i386/i386.opt: Likewise.
11559 * config/ia64/ia64.opt: Likewise.
11560 * config/rs6000/rs6000.opt: Likewise.
11561 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
11562 Remove usage of flag_check_pointer_bounds.
11563 * lto-wrapper.c (merge_and_complain): Do not handle
11564 OPT_fcheck_pointer_bounds.
11565 (append_compiler_options): Likewise.
11566 * opt-functions.awk: Do not handle Deprecated.
11567 * optc-gen.awk: Check that Var, Report and Init are not
11568 used for an option with Ignore/Deprecated flag.
11569 * opts-common.c (decode_cmdline_option): Do not report
11571 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
11573 * opts.h (struct cl_option): Remove cl_deprecated flag.
11574 (CL_ERR_DEPRECATED): Remove error enum value.
11576 2018-08-17 Richard Biener <rguenther@suse.de>
11578 PR middle-end/86505
11579 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
11580 across a va-arg-pack using call adjust its return value accordingly.
11582 2018-08-16 Martin Sebor <msebor@redhat.com>
11584 PR tree-optimization/86853
11585 * gimple-ssa-sprintf.c (struct format_result): Rename member.
11586 (struct fmtresult): Add member and initialize it in ctors.
11587 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
11588 (format_string): Handle %S the same as %ls. Set MAYFAIL.
11589 (format_directive): Set POSUNDER4K when MAYFAIL is set.
11590 (parse_directive): Handle %C same as %c.
11591 (sprintf_dom_walker::compute_format_length): Adjust.
11592 (is_call_safe): Adjust.
11594 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
11596 * builtins.c (c_strlen): Add new parameter eltsize. Use it
11597 for determining how to count the elements.
11598 * builtins.h (c_strlen): Adjust prototype.
11599 * expr.c (string_constant): Add new parameter mem_size.
11600 Set *mem_size appropriately.
11601 * expr.h (string_constant): Adjust protoype.
11602 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
11603 * gimple-fold.h (get_range_strlen): Adjust prototype.
11604 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
11605 (format_string): Call get_string_length with eltsize.
11607 2018-08-16 David Malcolm <dmalcolm@redhat.com>
11609 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
11610 to emit the span, rather than setting it as the prefix.
11612 2018-08-16 David Malcolm <dmalcolm@redhat.com>
11614 * diagnostic-show-locus.c (layout::start_annotation_line): Add
11615 "margin_char" parameter, defaulting to space. Use it in place
11616 of pp_space for the initial part of the margin.
11617 (layout::print_leading_fixits): Use '+' when filling the margin
11618 of line-insertion fix-it hints.
11620 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
11622 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
11625 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
11627 * config/rs6000/altivec.md: Don't set length attribute to the default
11629 * config/rs6000/darwin.md: Ditto.
11630 * config/rs6000/dfp.md: Ditto.
11631 * config/rs6000/htm.md: Ditto.
11632 * config/rs6000/rs6000.md: Ditto.
11633 * config/rs6000/sync.md: Ditto.
11634 * config/rs6000/vsx.md: Ditto.
11636 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
11638 * config/rs6000/altivec.md: Don't set length attribute to the default
11639 value, for branch instructions.
11640 * config/rs6000/darwin.md: Ditto.
11641 * config/rs6000/rs6000.md: Ditto.
11643 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
11645 * config/rs6000/rs6000.md (length): Always define as const_int 4.
11646 (unnamed conditional branch define_insn): Set length to 4 or 8
11647 depending on offset.
11648 (<bd>_<mode>): Similar, for alternative 0.
11649 (<bd>tf_<mode>): Ditto.
11651 2018-08-16 Tamar Christina <tamar.christina@arm.com>
11653 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
11655 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
11657 * doc/rtl.texi: Replace old RTX class names with new names.
11660 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
11662 * expmed.h (canonicalize_comparison): New declaration.
11663 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
11664 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
11665 * optabs.c (prepare_cmp_insn): Likewise.
11666 * rtl.h (unsigned_condition_p): New function which checks if a
11667 comparison operator is unsigned.
11669 2018-08-16 Nathan Sidwell <nathan@acm.org>
11671 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
11672 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
11674 2018-08-16 Tamar Christina <tamar.christina@arm.com>
11677 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
11678 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
11679 (mov<mov>): ..this and enable unconditionally.
11681 2018-08-16 Tamar Christina <tamar.christina@arm.com>
11683 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
11685 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
11687 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
11688 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
11689 "Common" with "Target".
11691 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
11693 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
11694 * doc/invoke.texi (mmitigate-rop): Remove.
11695 * config/i386/i386.c: Do not include "regrename.h".
11696 (ix86_rop_should_change_byte_p, reg_encoded_number)
11697 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
11699 (ix86_reorg): Remove call to ix86_mitigate_rop.
11700 * config/i386/i386.md (attr "modrm_class"): Remove.
11701 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
11702 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
11703 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
11705 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
11707 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
11708 allow folding of mergeh() and mergel() for the float and double types.
11709 (fold_mergehl_helper): Rework to handle building a permute tree
11712 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
11714 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
11717 2018-08-15 David Malcolm <dmalcolm@redhat.com>
11719 * common.opt (fdiagnostics-show-labels): New option.
11720 * diagnostic-show-locus.c (class layout_range): Add field
11722 (class layout): Add field "m_show_labels_p".
11723 (layout_range::layout_range): Add param "label" and use it to
11724 initialize m_label.
11725 (make_range): Pass in NULL for new "label" param of layout_range's
11727 (layout::layout): Initialize m_show_labels_p.
11728 (layout::maybe_add_location_range): Pass in loc_range->m_label
11729 when constructing layout_range instances.
11730 (struct line_label): New struct.
11731 (layout::print_any_labels): New member function.
11732 (layout::print_line): Call it if label-printing is enabled.
11733 (selftest::test_one_liner_labels): New test.
11734 (selftest::test_diagnostic_show_locus_one_liner): Call it.
11735 * diagnostic.c (diagnostic_initialize): Initialize
11736 context->show_labels_p.
11737 * diagnostic.h (struct diagnostic_context): Add field
11739 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11740 -fno-diagnostics-show-labels.
11741 * dwarf2out.c (gen_producer_string): Add
11742 OPT_fdiagnostics_show_labels to the ignored options.
11743 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
11745 (gcc_rich_location::maybe_add_expr): Likewise.
11746 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
11747 label" param, defaulting to NULL.
11748 (gcc_rich_location::add_expr): Add "label" param.
11749 (gcc_rich_location::maybe_add_expr): Likewise.
11750 (class text_range_label): New class.
11751 (class range_label_for_type_mismatch): New class.
11752 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
11753 of format_warning_va.
11754 (fmtwarn_n): Likewise for new params of format_warning_n_va.
11755 * lto-wrapper.c (merge_and_complain): Add
11756 OPT_fdiagnostics_show_labels to the "pick one setting" options.
11757 (append_compiler_options): Likewise to the dropped options.
11758 (append_diag_options): Likewise to the passed-on options.
11759 * opts.c (common_handle_option): Handle the new option.
11760 * selftest-diagnostic.c
11761 (test_diagnostic_context::test_diagnostic_context): Enable
11763 * substring-locations.c: Include "gcc-rich-location.h".
11764 (format_warning_n_va): Add "fmt_label" and "param_label" params
11765 and use them as appropriate.
11766 (format_warning_va): Add "fmt_label" and "param_label" params,
11767 passing them on to format_warning_n_va.
11768 (format_warning_at_substring): Likewise.
11769 (format_warning_at_substring_n): Likewise.
11770 * substring-locations.h (format_warning_va): Add "fmt_label" and
11771 "param_label" params.
11772 (format_warning_n_va): Likewise.
11773 (format_warning_at_substring): Likewise.
11774 (format_warning_at_substring_n): Likewise.
11775 * toplev.c (general_init): Initialize global_dc->show_labels_p.
11777 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
11780 * builtins.c (expand_builtin_memcmp): Do not expand the call
11781 when overflow is detected.
11783 2018-08-15 Martin Sebor <msebor@redhat.com>
11785 PR tree-optimization/71625
11786 * config/aarch64/aarch64-builtins.c
11787 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
11789 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
11791 * config/s390/s390.c (s390_reorg): Remove loop.
11793 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
11796 (darwin_function_switched_text_sections): Delete.
11797 * gcc/config/darwin.h
11798 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
11800 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
11803 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
11804 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
11805 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
11807 2018-08-15 Martin Liska <mliska@suse.cz>
11809 PR tree-optimization/86925
11810 * predict.c (expr_expected_value_1): When taking
11811 later predictor, assign also probability.
11812 Use fold_build2_initializer_loc in order to fold
11813 the expression in -frounding-math.
11815 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
11817 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
11819 (expand_vec_perm_1): Try the new method.
11821 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
11824 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
11825 Check whether lra_live_max_point is 0 before dividing.
11827 2018-08-14 Martin Sebor <msebor@redhat.com>
11829 PR tree-optimization/86650
11830 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
11831 (vrp_prop::check_mem_ref): Same.
11833 2018-08-13 Liu Hao <lh_mouse@126.com>
11835 * pretty-print.c (eat_esc_sequence): Swap the foreground and
11836 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
11837 and clear it thereafter, as it only works for DBCS.
11839 2018-08-13 Liu Hao <lh_mouse@126.com>
11841 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
11842 handle returned by _get_osf_handle().
11844 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11846 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
11847 for folding vec_perm.
11849 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11851 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
11852 Add support for gimple-folding of vec_pack() and vec_unpack()
11855 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11857 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
11858 vec_xst variants to the list.
11859 (rs6000_gimple_fold_builtin): Add support for folding unaligned
11860 vector loads and stores.
11862 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
11864 * config.gcc (rs6000-ibm-aix4.x): Delete.
11865 (rs6000-ibm-aix5.1): Delete.
11866 (rs6000-ibm-aix5.2): Delete.
11867 (rs6000-ibm-aix5.3): Delete.
11868 * config/rs6000/aix43.h: Delete.
11869 * config/rs6000/aix51.h: Delete.
11870 * config/rs6000/aix52.h: Delete.
11871 * config/rs6000/t-aix43: Delete.
11873 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
11875 * config/s390/s390.c (s390_decompose_constant_pool_ref):
11877 (s390_decompose_address): Factor out constant pool ref
11880 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
11882 * config/nds32/nds32-predicates.c
11883 (nds32_can_use_bclr_p): Change return type as bool.
11884 (nds32_can_use_bset_p): Ditto.
11885 (nds32_can_use_btgl_p): Ditto.
11886 (nds32_can_use_bitci_p): Ditto.
11887 * config/nds32/nds32-protos.h
11888 (nds32_can_use_bclr_p): Change declaration.
11889 (nds32_can_use_bset_p): Ditto.
11890 (nds32_can_use_btgl_p): Ditto.
11891 (nds32_can_use_bitci_p): Ditto.
11893 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
11895 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
11896 Support -msched-prolog-epilog option.
11897 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
11899 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
11901 * common/config/nds32/nds32-common.c
11902 (nds32_option_optimization_table): Enalbe -malways-align.
11904 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
11906 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
11908 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
11909 OPT_misr_secure_ case.
11910 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
11911 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
11912 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
11914 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
11915 (nds32_isr_info): New field security_level.
11916 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
11917 * config/nds32/nds32.md (return_internal): Consider critical attribute.
11918 * config/nds32/nds32.opt (misr-secure): New option.
11919 * config/nds32/nds32_init.inc: New file.
11920 * config/nds32/nds32_isr.h: New file.
11922 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
11924 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
11925 Update comment for atomic instructions.
11926 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
11927 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
11929 (atomic_loaddi): Revise fence expansion to only emit fence prior to
11930 load for __ATOMIC_SEQ_CST model.
11931 (atomic_loaddi_1): Remove float register target.
11932 (atomic_storedi): Handle CONST_INT values.
11933 (atomic_storedi_1): Remove float register source. Add special case
11935 (memory_barrier): New expander and insn.
11937 2018-08-11 Jakub Jelinek <jakub@redhat.com>
11939 PR tree-optimization/86835
11940 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
11941 new_stmt after def_gsi, make sure to insert new_square_stmt after
11942 that stmt, not 2 stmts before it.
11944 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
11947 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
11950 2018-08-10 Martin Liska <mliska@suse.cz>
11953 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
11955 * builtins.c (expand_builtin_expect_with_probability):
11957 (expand_builtin_expect_with_probability): New function.
11958 (build_builtin_expect_predicate): Add new argumnet probability
11959 for BUILT_IN_EXPECT_WITH_PROBABILITY.
11960 (fold_builtin_expect):
11963 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
11964 * builtins.h (fold_builtin_expect): Set new argument.
11965 * doc/extend.texi: Document __builtin_expect_with_probability.
11966 * doc/invoke.texi: Likewise.
11967 * gimple-fold.c (gimple_fold_call): Pass new argument.
11968 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
11969 also BUILT_IN_EXPECT_WITH_PROBABILITY.
11970 * predict.c (get_predictor_value): New function.
11971 (expr_expected_value): Add new argument probability. Assume
11972 that predictor and probability are always non-null.
11973 (expr_expected_value_1): Likewise. For __builtin_expect and
11974 __builtin_expect_with_probability set probability. Handle
11975 combination in binary expressions.
11976 (tree_predict_by_opcode): Simplify code by simply calling
11977 get_predictor_value.
11978 (pass_strip_predict_hints::execute): Add handling of
11979 BUILT_IN_EXPECT_WITH_PROBABILITY.
11980 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
11982 * tree.h (DECL_BUILT_IN_P): New function.
11984 2018-08-10 Martin Liska <mliska@suse.cz>
11986 PR tree-optimization/85799
11987 * passes.def: Add argument for pass_strip_predict_hints.
11988 * predict.c (class pass_strip_predict_hints): Add new argument
11990 (strip_predictor_early): New function.
11991 (pass_strip_predict_hints::execute): Call the function to
11993 (strip_predict_hints): New function.
11994 * predict.def: Fix comment.
11996 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
11998 * Makefile.in: Clarify which tm.texi to copy over to assert the
11999 right to grant a GFDL license for all.
12001 2018-08-09 Jeff Law <law@redhat.com>
12003 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
12006 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
12008 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
12011 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
12013 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
12014 reductions for variable-length vectors.
12016 2018-08-09 David Malcolm <dmalcolm@redhat.com>
12019 * common.opt (fdiagnostics-show-line-numbers): New option.
12020 * diagnostic-show-locus.c (class layout): Add fields
12021 "m_show_line_numbers_p" and "m_linenum_width";
12022 (num_digits): New function.
12023 (test_num_digits): New function.
12024 (layout::layout): Initialize new fields. Update m_x_offset
12025 logic to handle any left margin.
12026 (layout::print_source_line): Print line number when requested.
12027 (layout::start_annotation_line): New member function.
12028 (layout::print_annotation_line): Call it.
12029 (layout::print_leading_fixits): Likewise.
12030 (layout::print_trailing_fixits): Likewise. Update calls to
12031 move_to_column for new parameter.
12032 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
12033 it to potentially call start_annotation_line.
12034 (layout::show_ruler): Call start_annotation_line.
12035 (selftest::test_line_numbers_multiline_range): New selftest.
12036 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
12037 and selftest::test_line_numbers_multiline_range.
12038 * diagnostic.c (diagnostic_initialize): Initialize
12039 show_line_numbers_p.
12040 * diagnostic.h (struct diagnostic_context): Add field
12041 "show_line_numbers_p".
12042 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
12043 -fno-diagnostics-show-line-numbers.
12044 * dwarf2out.c (gen_producer_string): Add
12045 OPT_fdiagnostics_show_line_numbers to the ignored options.
12046 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
12047 one setting" options.
12048 (append_compiler_options): Likewise to the dropped options.
12049 (append_diag_options): Likewise to the passed-on options.
12050 * opts.c (common_handle_option): Handle the new option.
12051 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
12053 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
12055 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
12056 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
12057 third argument of type "const signed char" to descriptions of
12058 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
12059 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
12060 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
12061 __builtin_bcdsub_ov functions.
12063 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
12065 PR tree-optimization/86858
12066 * tree-vect-loop.c (vect_is_simple_reduction): Restore
12067 flow_bb_inside_loop_p calls.
12069 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
12071 PR tree-optimization/86871
12072 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
12073 instead of gimple_assign_lhs.
12075 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
12078 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
12079 register constraint to operand 0.
12080 (add<mode>3_carryinC): Likewise.
12081 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
12083 2018-08-09 Martin Liska <mliska@suse.cz>
12086 * common.opt: Remove extra line.
12088 2018-08-09 Martin Liska <mliska@suse.cz>
12090 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
12091 at the end of a line, make first letter capital and end up
12092 a sentence with a dot.
12093 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
12094 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
12095 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
12096 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
12097 (PARAM_MAX_ISL_OPERATIONS): Likewise.
12098 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
12099 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
12100 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
12101 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
12102 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
12103 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
12104 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
12105 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
12106 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
12107 (PARAM_TREE_REASSOC_WIDTH): Likewise.
12108 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
12109 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
12110 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
12112 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
12115 * config/s390/s390.c (s390_option_override_internal): Reduce the
12116 stack-clash-protection-probe-interval param if it would be too big
12119 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
12122 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
12123 * config/m68k/m68k.c (handle_move_double): Don't call
12124 m68k_final_prescan_insn.
12125 (m68k_adjust_decorated_operand): Renamed from
12126 m68k_final_prescan_insn, remove first and third operand and
12128 (print_operand): Call it.
12129 (print_operand_address): Call it.
12131 2018-08-08 Nathan Sidwell <nathan@acm.org>
12133 * diagnostic.c (diagnostic_report_current_module): Use
12134 linemap_included_from & linemap_included_from_linemap.
12136 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
12138 * config/aarch64/aarch64-cores.def: Add phecda core.
12139 * config/aarch64/aarch64-tune.md: Regenerate.
12140 * doc/invoke.texi: Add phecda core.
12142 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
12145 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
12147 * config/s390/s390.md ("movti"): Add more alternatives for
12148 constant to GPR copies.
12150 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
12152 * config/s390/s390.c: Fix whitespace damage throughout the file.
12153 * config/s390/s390.h: Likewise.
12154 * config/s390/tpf.h: Likewise.
12156 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
12158 * config/s390/s390.c (s390_loadrelative_operand_p):
12159 Remove TARGET_CPU_ZARCH usages.
12160 (s390_rtx_costs): Likewise.
12161 (s390_legitimate_constant_p): Likewise.
12162 (s390_cannot_force_const_mem): Likewise.
12163 (legitimate_reload_constant_p): Likewise.
12164 (s390_preferred_reload_class): Likewise.
12165 (legitimize_pic_address): Likewise.
12166 (legitimize_tls_address): Likewise.
12167 (s390_split_branches): Removed.
12168 (s390_add_execute): Removed.
12169 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
12170 (s390_mainpool_start): Likewise.
12171 (s390_mainpool_finish): Likewise.
12172 (s390_mainpool_cancel): Removed.
12173 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
12174 (s390_chunkify_cancel): Likewise.
12175 (s390_return_addr_rtx): Likewise.
12176 (s390_register_info): Remove split_branches_pending_p uages.
12177 (s390_optimize_register_info): Likewise.
12178 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
12179 split_branches_pending_p usages.
12180 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
12181 (s390_load_got): Likewise.
12182 (s390_expand_split_stack_prologue): Likewise.
12183 (output_asm_nops): Likewise.
12184 (s390_function_profiler): Likewise.
12185 (s390_emit_call): Likewise.
12186 (s390_conditional_register_usage): Likewise.
12187 (s390_optimize_prologue): Likewise.
12188 (s390_reorg): Remove TARGET_CPU_ZARCH and
12189 split_branches_pending_p usages.
12190 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
12192 (s390_output_indirect_thunk_function): Likewise.
12193 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
12194 (TARGET_CPU_ZARCH_P): Removed.
12195 (struct machine_function): Remove split_branches_pending_p.
12196 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
12198 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
12200 * common/config/s390/s390-common.c (processor_flags_table):
12202 * config.gcc: Remove with_arch/with_tune support.
12203 * config/s390/2064.md: Remove cpu attribute comparisons.
12204 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12206 * config/s390/linux.h (ASM_SPEC):
12207 Remove -march support.
12208 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
12209 Use a table to get an arch level.
12210 * config/s390/s390-opts.h (enum processor_type):
12211 Remove enum values.
12212 * config/s390/s390.c
12213 (processor_table): Remove entries, add arch_level values.
12214 (s390_issue_rate): Remove cases.
12215 (s390_option_override): Adjust
12216 s390_option_override_internal() call.
12217 (s390_option_override_internal): Remove deprecation warning.
12218 (s390_valid_target_attribute_tree): Adjust
12219 s390_option_override_internal() call.
12220 * config/s390/s390.h (struct s390_processor):
12221 Share with s390-c.c, add arch_level field.
12222 * config/s390/s390.md:
12223 Remove occurrences in cpu attribute.
12224 * config/s390/s390.opt: Remove -march/-mtune support.
12225 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
12226 * doc/invoke.texi: Remove deprecation warning.
12228 2018-08-08 Luis Machado <luis.machado@linaro.org>
12230 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
12232 (qdf24xx_tunings): Set vector cost structure to
12233 qdf24xx_vector_cost.
12235 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
12236 <register_sextend>: Set to 3.
12238 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
12241 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
12242 * config/aarch64/aarch64-simd.md
12243 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
12244 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
12245 (aarch64_frecpx<mode>): ...this new pattern.
12246 * config/aarch64/aarch64-simd-builtins.def: Remove comment
12247 about aarch64_frecp<FRECP:frecp_suffix><mode>.
12249 2018-08-07 Martin Liska <mliska@suse.cz>
12251 PR middle-end/83023
12252 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
12253 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
12254 * predict.def (PRED_MALLOC_NONNULL): New predictor.
12255 * doc/extend.texi: Document that malloc attribute adds
12258 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
12261 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12262 Define to speculation_safe_value_not_needed.
12264 2018-08-06 Jeff Law <law@redhat.com>
12266 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
12267 the vr_values instance to cprop_into_stmt.
12268 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
12269 (cprop_operand): Also query EVRP to determine if OP is a constant.
12271 2018-08-06 Nathan Sidwell <nathan@acm.org>
12273 * diagnostic.c (diagnostic_report_current_module): Reroll
12274 included-at loop. Translate text.
12276 2018-08-06 David Malcolm <dmalcolm@redhat.com>
12278 * function-tests.c (selftest::test_expansion_to_rtl): Call
12279 free_after_compilation.
12281 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12283 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
12285 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
12287 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
12288 loops with memory block operations from getting unrolled.
12290 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
12293 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12294 Define to speculation_safe_value_not_needed.
12296 2018-08-06 Jeff Law <law@redhat.com>
12298 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
12301 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
12304 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
12305 with all enabled __intN types.
12307 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
12309 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12311 * alias.c (record_set): Check for clobber high.
12312 * cfgexpand.c (expand_gimple_stmt): Likewise.
12313 * combine-stack-adj.c (single_set_for_csa): Likewise.
12314 * combine.c (find_single_use_1): Likewise.
12315 (set_nonzero_bits_and_sign_copies): Likewise.
12316 (get_combine_src_dest): Likewise.
12317 (is_parallel_of_n_reg_sets): Likewise.
12318 (try_combine): Likewise.
12319 (record_dead_and_set_regs_1): Likewise.
12320 (reg_dead_at_p_1): Likewise.
12321 (reg_dead_at_p): Likewise.
12322 * dce.c (deletable_insn_p): Likewise.
12323 (mark_nonreg_stores_1): Likewise.
12324 (mark_nonreg_stores_2): Likewise.
12325 * df-scan.c (df_find_hard_reg_defs): Likewise.
12326 (df_uses_record): Likewise.
12327 (df_get_call_refs): Likewise.
12328 * dwarf2out.c (mem_loc_descriptor): Likewise.
12329 * haifa-sched.c (haifa_classify_rtx): Likewise.
12330 * ira-build.c (create_insn_allocnos): Likewise.
12331 * ira-costs.c (scan_one_insn): Likewise.
12332 * ira.c (equiv_init_movable_p): Likewise.
12333 (rtx_moveable_p): Likewise.
12334 (interesting_dest_for_shprep): Likewise.
12335 * jump.c (mark_jump_label_1): Likewise.
12336 * postreload-gcse.c (record_opr_changes): Likewise.
12337 * postreload.c (reload_cse_simplify): Likewise.
12338 (struct reg_use): Add source expr.
12339 (reload_combine): Check for clobber high.
12340 (reload_combine_note_use): Likewise.
12341 (reload_cse_move2add): Likewise.
12342 (move2add_note_store): Likewise.
12343 * print-rtl.c (print_pattern): Likewise.
12344 * recog.c (decode_asm_operands): Likewise.
12345 (store_data_bypass_p): Likewise.
12346 (if_test_bypass_p): Likewise.
12347 * regcprop.c (kill_clobbered_value): Likewise.
12348 (kill_set_value): Likewise.
12349 * reginfo.c (reg_scan_mark_refs): Likewise.
12350 * reload1.c (maybe_fix_stack_asms): Likewise.
12351 (eliminate_regs_1): Likewise.
12352 (elimination_effects): Likewise.
12353 (mark_not_eliminable): Likewise.
12354 (scan_paradoxical_subregs): Likewise.
12355 (forget_old_reloads_1): Likewise.
12356 * reorg.c (find_end_label): Likewise.
12357 (try_merge_delay_insns): Likewise.
12358 (redundant_insn): Likewise.
12359 (own_thread_p): Likewise.
12360 (fill_simple_delay_slots): Likewise.
12361 (fill_slots_from_thread): Likewise.
12362 (dbr_schedule): Likewise.
12363 * resource.c (update_live_status): Likewise.
12364 (mark_referenced_resources): Likewise.
12365 (mark_set_resources): Likewise.
12366 * rtl.c (copy_rtx): Likewise.
12367 * rtlanal.c (reg_referenced_p): Likewise.
12368 (single_set_2): Likewise.
12369 (noop_move_p): Likewise.
12370 (note_stores): Likewise.
12371 * sched-deps.c (sched_analyze_reg): Likewise.
12372 (sched_analyze_insn): Likewise.
12374 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12376 * cse.c (invalidate_reg): New function extracted from...
12377 (invalidate): ...here.
12378 (canonicalize_insn): Check for clobber high.
12379 (invalidate_from_clobbers): invalidate clobber highs.
12380 (invalidate_from_sets_and_clobbers): Likewise.
12381 (count_reg_usage): Check for clobber high.
12382 (insn_live_p): Likewise.
12383 * cselib.c (cselib_expand_value_rtx_1):Likewise.
12384 (cselib_invalidate_regno): Check for clobber in setter.
12385 (cselib_invalidate_rtx): Pass through setter.
12386 (cselib_invalidate_rtx_note_stores):
12387 (cselib_process_insn): Check for clobber high.
12388 * cselib.h (cselib_invalidate_rtx): Add operand.
12390 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12392 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
12393 (mark_not_eliminable): Likewise.
12394 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
12395 * lra-lives.c (process_bb_lives): Check for clobber high.
12396 * lra.c (new_insn_reg): Remember clobber highs.
12397 (collect_non_operand_hard_regs): Check for clobber high.
12398 (lra_set_insn_recog_data): Likewise.
12399 (add_regs_to_insn_regno_info): Likewise.
12400 (lra_update_insn_regno_info): Likewise.
12402 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12404 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
12405 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
12407 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12409 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
12410 (copy_insn_1): Likewise.
12411 (gen_hard_reg_clobber_high): New gen function.
12412 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
12413 * genemit.c (gen_exp): Likewise.
12414 (gen_emit_seq): Pass through info.
12415 (gen_insn): Check for CLOBBER_HIGH.
12416 (gen_expand): Pass through info.
12417 (gen_split): Likewise.
12418 (output_add_clobbers): Likewise.
12419 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
12420 (remove_clobbers): Likewise.
12421 * rtl.h (gen_hard_reg_clobber_high): New declaration.
12423 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12425 * doc/rtl.texi (clobber_high): Add.
12426 (parallel): Add in clobber high
12427 * rtl.c (rtl_check_failed_code3): Add function.
12428 * rtl.def (CLOBBER_HIGH): Add expression.
12429 * rtl.h (RTL_CHECKC3): Add macro.
12430 (rtl_check_failed_code3): Add declaration.
12431 (XC3EXP): Add macro.
12433 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
12436 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
12437 cfun->machine->max_used_stack_alignment if needed.
12439 2018-08-04 Martin Sebor <msebor@redhat.com>
12441 PR tree-optimization/86571
12442 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
12445 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
12448 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12451 2018-08-03 Jeff Law <law@redhat.com>
12454 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12455 Define to speculation_safe_value_not_needed.
12457 2018-08-03 David Malcolm <dmalcolm@redhat.com>
12459 * doc/gcov.texi (-x): Remove duplicate "to".
12460 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
12461 (-Wif-not-aligned): Remove duplicate "is".
12462 (-flto): Remove duplicate "the".
12463 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
12464 duplicate "v5.00.b".
12465 (MSP430 Options): Remove duplicate "and" from the description
12466 of "-mgprel-sec=regexp".
12467 (x86 Options): Remove duplicate copies of "vmldLog102" and
12468 vmlsLog104 from description of "-mveclibabi=type".
12470 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
12472 * internal-fn.h (first_commutative_argument): Declare.
12473 * internal-fn.c (first_commutative_argument): New function.
12474 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
12475 restrictions for pattern statements. Use first_commutative_argument
12476 to look for commutative operands in calls to internal functions.
12478 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
12480 * Makefile.in (wide-int-range.o): New.
12481 * tree-vrp.c: Move all the wide_int_* functions to...
12482 * wide-int-range.cc: ...here.
12483 * tree-vrp.h: Move all the wide_int_* prototypes to...
12484 * wide-int-range.h: ...here.
12486 2018-08-03 Tom de Vries <tdevries@suse.de>
12488 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
12490 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
12491 * except.c (output_function_exception_table): Do early exit if
12492 targetm_common.except_unwind_info (&global_options) == UI_NONE.
12494 2018-08-03 Martin Liska <mliska@suse.cz>
12496 * predict.c (dump_prediction): Change to 2 digits
12499 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
12501 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
12503 (falkor_am_1_gtov_gtov): ... a new insn reservation.
12505 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
12507 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
12508 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
12509 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
12510 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
12511 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
12513 2018-08-02 David Malcolm <dmalcolm@redhat.com>
12515 * diagnostic-show-locus.c (diagnostic_show_locus): Use
12516 pp_take_prefix when saving the existing prefix.
12517 * diagnostic.c (diagnostic_append_note): Likewise.
12518 * langhooks.c (lhd_print_error_function): Likewise.
12519 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
12520 param's type. Free the existing prefix.
12521 (pp_take_prefix): New function.
12522 (pretty_printer::pretty_printer): Drop the prefix parameter.
12523 Rename the length parameter to match the comment.
12524 (pretty_printer::~pretty_printer): Free the prefix.
12525 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
12527 (struct pretty_printer): Drop the "const" from "prefix" field's
12528 type and clarify memory management.
12529 (pp_set_prefix): Drop the "const" from the 2nd param.
12530 (pp_take_prefix): New decl.
12532 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
12534 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
12535 for word_mode_ok here instead of passing as argument.
12536 (expand_block_compare): Change select_block_compare_mode() call.
12537 (expand_strncmp_gpr_sequence): New function.
12538 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
12540 2018-08-02 Jeff Law <law@redhat.com>
12543 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12544 Define to speculation_safe_value_not_needed.
12547 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12548 Define to speculation_safe_value_not_needed.
12550 2018-08-02 Tom de Vries <tdevries@suse.de>
12553 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
12554 function. Return UI_TARGET unconditionally.
12555 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
12556 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
12558 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
12560 * genemit.c (print_overload_test): Fix typo.
12562 2018-08-02 Richard Biener <rguenther@suse.de>
12564 PR tree-optimization/86816
12565 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
12566 which checks for value availability before querying it.
12567 (gvn_uses_equal): Use it.
12568 (same_succ_hash): Likewise.
12569 (gimple_equal_p): Likewise.
12571 2018-08-02 Nick Clifton <nickc@redhat.com>
12574 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12575 Define to speculation_safe_value_not_needed.
12578 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12579 Define to speculation_safe_value_not_needed.
12582 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12583 Define to speculation_safe_value_not_needed.
12586 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12587 Define to speculation_safe_value_not_needed.
12590 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12591 Define to speculation_safe_value_not_needed.
12594 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12595 Define to speculation_safe_value_not_needed.
12598 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12599 Define to speculation_safe_value_not_needed.
12602 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12603 Define to speculation_safe_value_not_needed.
12606 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
12607 speculation_safe_value_not_needed.
12610 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
12611 to speculation_safe_value_not_needed.
12613 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
12615 * doc/md.texi: Expand the documentation of instruction names
12616 to mention port-local uses. Document '@' in pattern names.
12617 * read-md.h (overloaded_instance, overloaded_name): New structs.
12618 (mapping): Declare.
12619 (md_reader::handle_overloaded_name): New member function.
12620 (md_reader::get_overloads): Likewise.
12621 (md_reader::m_first_overload): New member variable.
12622 (md_reader::m_next_overload_ptr): Likewise.
12623 (md_reader::m_overloads_htab): Likewise.
12624 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
12625 m_next_overload_ptr and m_overloads_htab.
12626 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
12627 (get_mode_token, get_code_token, get_int_token): New functions.
12628 (map_attr_string): Add an optional argument that passes back
12629 the associated iterator.
12630 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
12631 (md_reader::handle_overloaded_name, add_overload_instance): New
12633 (apply_iterators): Handle '@' names. Report an error if '@'
12634 is used without iterators.
12635 (initialize_iterators): Initialize the new iterator_group fields.
12636 * genopinit.c (handle_overloaded_code_for)
12637 (handle_overloaded_gen): New functions.
12638 (main): Use them to print declarations of maybe_code_for_* and
12639 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
12640 * genemit.c (print_overload_arguments, print_overload_test)
12641 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
12642 (main): Use it to print definitions of maybe_code_for_* and
12643 maybe_gen_* functions.
12644 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
12645 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
12646 instead of explicit mode checks.
12647 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
12648 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
12649 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
12650 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
12651 (aarch64_expand_compare_and_swap): Likewise
12652 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
12653 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
12654 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
12655 (aarch64_constant_pool_reload_icode): Delete.
12656 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
12657 instead of aarch64_constant_pool_reload_icode. Use
12658 code_for_aarch64_reload_mov instead of explicit mode checks.
12659 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
12660 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
12661 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
12662 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
12663 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
12664 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
12665 (aarch64_atomic_load_op_code): Delete.
12666 (aarch64_emit_atomic_load_op): Likewise.
12667 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
12668 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
12669 instead of aarch64_emit_atomic_load_op.
12670 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
12671 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
12672 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
12673 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
12674 character before the pattern name.
12675 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
12676 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
12677 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
12678 (aarch64_frecps<mode>): Likewise.
12679 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
12680 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
12681 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
12682 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
12683 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
12685 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
12687 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
12688 Allow HFmode constants if TARGET_FP_F16INST.
12690 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
12693 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
12694 No longer check last store for clobber of address register.
12696 2018-08-02 Martin Liska <mliska@suse.cz>
12698 PR gcov-profile/86817
12699 * gcov.c (process_all_functions): New function.
12701 (process_file): Move functions processing to
12702 process_all_functions.
12704 2018-08-02 David Malcolm <dmalcolm@redhat.com>
12706 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
12707 "const" to the "gimple *" and "rtx_insn *" parameters.
12708 * dumpfile.h (dump_user_location_t::dump_user_location_t):
12710 (dump_location_t::dump_location_t): Likewise.
12712 2018-08-01 Martin Sebor <msebor@redhat.com>
12714 PR tree-optimization/86650
12715 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
12716 rather than a "gcall *". Directly pass the data of interest
12717 to percent_K_format, rather than building a temporary CALL_EXPR
12719 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
12720 (gimple_fold_builtin_strncat): Adjust.
12721 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
12722 gcall* argument with gimple*.
12723 * gimple-ssa-warn-restrict.c (check_call): Same.
12724 (wrestrict_dom_walker::before_dom_children): Same.
12725 (builtin_access::builtin_access): Same.
12726 (check_bounds_or_overlap): Same
12727 (maybe_diag_overlap): Same.
12728 (maybe_diag_offset_bounds): Same.
12729 * tree-diagnostic.c (default_tree_printer): Move usage of
12730 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
12732 * tree-pretty-print.c (percent_K_format): Add argument.
12733 * tree-pretty-print.h: Add argument.
12734 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
12735 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
12736 (maybe_diag_stxncpy_trunc): Same.
12737 (handle_builtin_stxncpy): Same.
12738 (handle_builtin_strcat): Same.
12740 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12742 * match.pd: Optimise pointer range checks.
12744 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12746 PR tree-optimization/86758
12747 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
12748 to remove pattern statements.
12750 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12752 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
12753 result of dfs_enumerate_from when constructing stmt_vec_infos,
12754 instead of additionally calling get_loop_body.
12756 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12758 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
12760 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
12761 When creating an iv, assert that the step is not known to be zero.
12762 (vect_setup_realignment): Update call accordingly.
12763 * tree-vect-stmts.c (vectorizable_store): Likewise.
12764 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
12766 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12768 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
12769 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
12770 (vectorizable_reduction): Likewise.
12771 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
12772 (vect_detect_hybrid_slp_stmts): Likewise.
12773 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
12775 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
12777 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
12778 (wide_int_set_zero_nonzero_bits): ...this.
12779 (zero_nonzero_bits_from_vr): Rename to...
12780 (vrp_set_zero_nonzero_bits): ...this.
12781 (extract_range_from_multiplicative_op_1): Abstract wide int
12783 (wide_int_range_multiplicative_op): ...here.
12784 (extract_range_from_binary_expr_1): Extract wide int binary
12785 operations into their own functions.
12786 (wide_int_range_lshift): New.
12787 (wide_int_range_can_optimize_bit_op): New.
12788 (wide_int_range_shift_undefined_p): New.
12789 (wide_int_range_bit_xor): New.
12790 (wide_int_range_bit_ior): New.
12791 (wide_int_range_bit_and): New.
12792 (wide_int_range_trunc_mod): New.
12793 (extract_range_into_wide_ints): New.
12794 (vrp_shift_undefined_p): New.
12795 (extract_range_from_multiplicative_op): New.
12796 (vrp_can_optimize_bit_op): New.
12797 * tree-vrp.h (value_range::dump): New.
12798 (wide_int_range_multiplicative_op): New.
12799 (wide_int_range_lshift):New.
12800 (wide_int_range_shift_undefined_p): New.
12801 (wide_int_range_bit_xor): New.
12802 (wide_int_range_bit_ior): New.
12803 (wide_int_range_bit_and): New.
12804 (wide_int_range_trunc_mod): New.
12805 (zero_nonzero_bits_from_bounds): Rename to...
12806 (wide_int_set_zero_nonzero_bits): ...this.
12807 (zero_nonzero_bits_from_vr): Rename to...
12808 (vrp_set_zero_nonzero_bits): ...this.
12809 (range_easy_mask_min_max): Rename to...
12810 (wide_int_range_can_optimize_bit_op): this.
12811 * vr-values.c (simplify_bit_ops_using_ranges): Rename
12812 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
12814 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12816 * tree-vectorizer.h (vect_orig_stmt): New function.
12817 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
12818 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
12819 (vect_create_epilog_for_reduction): Likewise.
12820 (vectorizable_live_operation): Likewise.
12821 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
12822 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
12823 * tree-vect-stmts.c (vectorizable_call): Likewise.
12824 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
12826 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12828 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
12830 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
12831 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
12832 (vect_transform_loop): Likewise.
12833 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12835 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12837 * tree-vectorizer.h (vect_schedule_slp): Return void.
12838 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12839 (vect_schedule_slp): Likewise.
12841 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12843 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
12845 (vect_transform_loop): Update calls accordingly. Schedule SLP
12846 instances before the main loop, if any exist.
12848 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12850 PR tree-optimization/86749
12851 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
12852 If the lhs is used in a COND_EXPR, check that it is being used
12853 as the "then" or "else" value.
12855 2018-08-01 Tom de Vries <tdevries@suse.de>
12858 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
12859 speculation_safe_value_not_needed.
12861 2018-08-01 Richard Biener <rguenther@suse.de>
12863 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
12864 as base and offset.
12866 2018-08-01 Martin Liska <mliska@suse.cz>
12868 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
12869 format how successful transformation is dumped.
12870 (gimple_mod_pow2_value_transform): Likewise.
12871 (gimple_mod_subtract_transform): Likewise.
12872 (gimple_stringops_transform): Likewise.
12874 2018-08-01 Martin Liska <mliska@suse.cz>
12876 PR value-prof/35543
12877 * value-prof.c (interesting_stringop_to_profile_p):
12878 Simplify the code and add BUILT_IN_MEMMOVE.
12879 (gimple_stringops_transform): Likewise.
12881 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
12883 * config/aarch64/aarch64-simd.md
12884 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
12885 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
12886 use GPI iterator instead of SI mode.
12888 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
12890 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
12891 rs6000_speculation_barrier.
12892 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
12893 new barrier pattern name.
12895 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
12897 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
12898 (speculation_barrier): New insn.
12900 2018-08-01 Richard Biener <rguenther@suse.de>
12903 * graphite.h: Include isl/id.h and isl/space.h to allow build
12906 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
12909 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
12910 mode for COFF targets.
12911 * defaults.h (TARGET_COFF): Define.
12912 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
12913 TARGET_COFF): Define.
12914 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
12915 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
12916 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
12918 2018-07-31 Alexandre Oliva <oliva@adacore.com>
12920 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
12921 * gimple-streamer-out.c (output_bb): Save it.
12922 * lto-streamer-in.c (input_struct_function_base): Restore
12923 instance discriminator if available. Create map on demand.
12924 * lto-streamer-out.c (output_struct_function_base): Save it if
12926 * final.c (decl_to_instance_map): Document LTO strategy.
12928 2018-07-31 Alexandre Oliva <oliva@adacore.com>
12929 Olivier Hainque <hainque@adacore.com>
12931 * debug.h (decl_to_instance_map_t): New type.
12932 (decl_to_instance_map): Declare.
12933 (maybe_create_decl_to_instance_map): New inline function.
12934 * final.c (bb_discriminator, last_bb_discriminator): New statics,
12935 to track basic block discriminators.
12936 (final_start_function_1): Initialize them.
12937 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
12939 (decl_to_instance_map): New variable.
12940 (map_decl_to_instance, maybe_set_discriminator): New functions.
12941 (notice_source_line): Set discriminator.
12943 2018-07-31 Ian Lance Taylor <iant@golang.org>
12945 * targhooks.c (default_have_speculation_safe_value): Add
12948 2018-07-31 David Malcolm <dmalcolm@redhat.com>
12950 * dump-context.h: Include "pretty-print.h".
12951 (dump_context::refresh_dumps_are_enabled): New decl.
12952 (dump_context::emit_item): New decl.
12953 (class dump_context): Add fields "m_test_pp" and
12955 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
12956 (temp_dump_context::get_dumped_text): New decl.
12957 (class temp_dump_context): Add field "m_pp".
12958 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
12959 (dump_context::refresh_dumps_are_enabled): ...and add a test for
12961 (set_dump_file): Update for above change.
12962 (set_alt_dump_file): Likewise.
12963 (dump_loc): New overload, taking a pretty_printer *.
12964 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
12965 to any test pretty-printer.
12966 (make_item_for_dump_gimple_stmt): New function, adapted from
12967 optinfo::add_gimple_stmt.
12968 (dump_context::dump_gimple_stmt): Call it, and use the result,
12969 eliminating the direct usage of dump_file and alt_dump_file in
12970 favor of indirectly using them via emit_item.
12971 (make_item_for_dump_gimple_expr): New function, adapted from
12972 optinfo::add_gimple_expr.
12973 (dump_context::dump_gimple_expr): Call it, and use the result,
12974 eliminating the direct usage of dump_file and alt_dump_file in
12975 favor of indirectly using them via emit_item.
12976 (make_item_for_dump_generic_expr): New function, adapted from
12978 (dump_context::dump_generic_expr): Call it, and use the result,
12979 eliminating the direct usage of dump_file and alt_dump_file in
12980 favor of indirectly using them via emit_item.
12981 (make_item_for_dump_printf_va): New function, adapted from
12982 optinfo::add_printf_va.
12983 (make_item_for_dump_printf): New function.
12984 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
12985 and use the result, eliminating the direct usage of dump_file and
12986 alt_dump_file in favor of indirectly using them via emit_item.
12987 (make_item_for_dump_dec): New function.
12988 (dump_context::dump_dec): Call it, and use the result,
12989 eliminating the direct usage of dump_file and alt_dump_file in
12990 favor of indirectly using them via emit_item.
12991 (make_item_for_dump_symtab_node): New function, adapted from
12992 optinfo::add_symtab_node.
12993 (dump_context::dump_symtab_node): Call it, and use the result,
12994 eliminating the direct usage of dump_file and alt_dump_file in
12995 favor of indirectly using them via emit_item.
12996 (dump_context::begin_scope): Reimplement, avoiding direct usage
12997 of dump_file and alt_dump_file in favor of indirectly using them
12999 (dump_context::emit_item): New member function.
13000 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
13001 Set up test pretty-printer on the underlying context. Call
13002 refresh_dumps_are_enabled.
13003 (temp_dump_context::~temp_dump_context): Call
13004 refresh_dumps_are_enabled.
13005 (temp_dump_context::get_dumped_text): New member function.
13006 (selftest::verify_dumped_text): New function.
13007 (ASSERT_DUMPED_TEXT_EQ): New macro.
13008 (selftest::test_capture_of_dump_calls): Run all tests twice, with
13009 and then without optinfo enabled. Add uses of
13010 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
13011 * dumpfile.h: Update comment for the dump_* API.
13012 * optinfo-emit-json.cc
13013 (selftest::test_building_json_from_dump_calls): Update for new
13014 param for temp_dump_context ctor.
13015 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
13016 and "m_owned" field.
13017 (optinfo_item::~optinfo_item): Likewise.
13018 (optinfo::add_item): New member function.
13019 (optinfo::emit): Update comment.
13020 (optinfo::add_string): Delete.
13021 (optinfo::add_printf): Delete.
13022 (optinfo::add_printf_va): Delete.
13023 (optinfo::add_gimple_stmt): Delete.
13024 (optinfo::add_gimple_expr): Delete.
13025 (optinfo::add_tree): Delete.
13026 (optinfo::add_symtab_node): Delete.
13027 (optinfo::add_dec): Delete.
13028 * optinfo.h (class dump_context): New forward decl.
13029 (optinfo::add_item): New decl.
13030 (optinfo::add_string): Delete.
13031 (optinfo::add_printf): Delete.
13032 (optinfo::add_printf_va): Delete.
13033 (optinfo::add_gimple_stmt): Delete.
13034 (optinfo::add_gimple_expr): Delete.
13035 (optinfo::add_tree): Delete.
13036 (optinfo::add_symtab_node): Delete.
13037 (optinfo::add_dec): Delete.
13038 (optinfo::add_poly_int): Delete.
13039 (optinfo_item::optinfo_item): Remove "owned" param.
13040 (class optinfo_item): Remove field "m_owned".
13042 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
13044 PR middle-end/86705
13045 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
13046 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
13047 requested variable alignment.
13048 (expand_one_ssa_partition): Likewise.
13049 (expand_one_var): Likewise.
13051 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
13053 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
13054 to speculation_safe_value_not_needed.
13056 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
13058 * targhooks.h (speculation_safe_value_not_needed): New prototype.
13059 * targhooks.c (speculation_safe_value_not_needed): New function.
13060 * target.def (have_speculation_safe_value): Update documentation.
13061 * doc/tm.texi: Regenerated.
13063 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
13065 * config/aarch64/iterators.md (ALLI_TI): New iterator.
13066 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
13068 (despeculate_copy<ALLI:mode>_insn): New insn.
13069 (despeculate_copyti_insn): New insn.
13070 (despeculate_simple<ALLI:mode>): New insn
13071 (despeculate_simpleti): New insn.
13072 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
13074 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
13075 aarch64_speculation_safe_value.
13076 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
13078 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
13080 * config/aarch64/aarch64-speculation.cc: New file.
13081 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
13082 before pass_reorder_blocks.
13083 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
13085 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
13086 X14 and X15 when tracking speculation.
13087 * config/aarch64/aarch64.md (register name constants): Add
13088 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
13089 (unspec): Add UNSPEC_SPECULATION_TRACKER.
13090 (speculation_barrier): New insn attribute.
13091 (cmp<mode>): Allow SP in comparisons.
13092 (speculation_tracker): New insn.
13093 (speculation_barrier): Add speculation_barrier attribute.
13094 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
13095 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
13096 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
13098 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
13100 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
13101 aarch64_track_speculation is true.
13102 (tb<optab><mode>1): Likewise.
13103 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
13104 generate CB[N]Z when tracking speculation.
13105 (aarch64_split_compare_and_swap): Likewise.
13106 (aarch64_split_atomic_op): Likewise.
13108 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
13110 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
13112 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
13114 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
13115 (speculation_barrier): New insn.
13117 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
13119 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
13120 * config/arm/arm.md (speculation_barrier): New expand.
13121 (speculation_barrier_insn): New pattern.
13123 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
13125 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
13126 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
13127 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
13128 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
13130 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
13131 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
13132 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
13133 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
13134 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
13135 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
13136 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
13137 * builtins.c (expand_speculation_safe_value): New function.
13138 (expand_builtin): Call it.
13139 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
13140 * doc/extend.texi: Document __builtin_speculation_safe_value.
13141 * doc/md.texi: Document "speculation_barrier" pattern.
13142 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
13143 TARGET_HAVE_SPECULATION_SAFE_VALUE.
13144 * doc/tm.texi: Regenerated.
13145 * target.def (have_speculation_safe_value, speculation_safe_value): New
13147 * targhooks.c (default_have_speculation_safe_value): New function.
13148 (default_speculation_safe_value): New function.
13149 * targhooks.h (default_have_speculation_safe_value): Add prototype.
13150 (default_speculation_safe_value): Add prototype.
13152 2018-07-31 David Malcolm <dmalcolm@redhat.com>
13154 * dump-context.h (dump_context::dump_loc): New decl.
13155 * dumpfile.c (dump_context::dump_loc): New member function.
13156 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
13157 and dump_gimple_stmt.
13158 (dump_context::dump_gimple_expr_loc): Likewise, using
13160 (dump_context::dump_generic_expr_loc): Likewise, using
13162 (dump_context::dump_printf_loc_va): Likewise, using
13164 (dump_context::begin_scope): Explicitly using the global function
13165 "dump_loc", rather than the member function.
13167 2018-07-31 Martin Sebor <msebor@redhat.com>
13169 PR tree-optimization/86741
13170 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
13172 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
13174 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
13175 depend on whether prefetch instructions will be emitted or not.
13176 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
13177 will be emitted or not.
13178 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
13179 (TARGET_SETMEM_PFD): New macros.
13181 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13183 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
13184 (NULL_STMT_VEC_INFO): Delete.
13185 (stmt_vec_info::operator*): Likewise.
13186 (stmt_vec_info::operator gimple *): Likewise.
13187 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
13188 of NULL_STMT_VEC_INFO.
13189 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
13190 (vect_reassociating_reduction_p): Likewise.
13191 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
13192 (vectorizable_store): Likewise.
13193 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
13194 (vec_info::free_stmt_vec_infos): Likewise.
13196 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13198 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
13199 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
13201 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13203 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
13204 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
13205 (vec_info::free_stmt_vec_info): New private member functions.
13206 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
13207 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
13208 * tree-parloops.c (gather_scalar_reductions): Remove calls to
13209 set_stmt_vec_info_vec and free_stmt_vec_infos.
13210 * tree-vect-loop.c (_loop_vec_info): Remove call to
13211 set_stmt_vec_info_vec.
13212 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
13213 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
13214 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
13215 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
13216 (vec_info::free_stmt_vec_info): ...these new functions. Remove
13217 assignments in {vec_info::,}new_stmt_vec_info that are redundant
13218 with the clearing in the xcalloc.
13219 (stmt_vec_info_vec): Delete.
13220 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
13221 (vectorize_loops): Likewise.
13222 (vec_info::~vec_info): Remove argument from call to
13223 free_stmt_vec_infos.
13224 (vec_info::add_stmt): Remove vinfo argument from call to
13227 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13229 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
13230 rather than a gimple stmt.
13231 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
13232 information for pattern statements when passed the original
13233 statement; instead wait to be passed the pattern statement itself.
13234 Don't call set_vinfo_for_stmt here.
13235 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
13236 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
13237 stmt_vec_infos here.
13238 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
13239 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
13240 stmt_vec_infos entry.
13242 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13244 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
13245 * tree-vectorizer.c (vec_info::replace_stmt): New function.
13246 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
13247 * tree-vect-stmts.c (vectorizable_call): Likewise.
13248 (vectorizable_simd_clone_call): Likewise.
13250 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13252 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
13253 * tree-vectorizer.c (vec_info::remove_stmt): New function.
13254 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
13255 * tree-vect-loop.c (vect_transform_loop): Likewise.
13256 * tree-vect-slp.c (vect_schedule_slp): Likewise.
13257 * tree-vect-stmts.c (vect_remove_stores): Likewise.
13259 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13261 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
13262 (vect_dr_stmt): Delete.
13263 * tree-vectorizer.c (vec_info::lookup_dr): New function.
13264 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
13266 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
13267 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
13268 (vect_verify_datarefs_alignment, vect_peeling_supportable)
13269 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
13270 (vect_analyze_data_refs): Likewise.
13271 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
13273 (vect_find_same_alignment_drs): Likewise.
13274 (vect_slp_analyze_node_dependences): Update calls accordingly.
13275 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
13276 instead of DR_VECT_AUX.
13277 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
13278 of a vector data references. Use vec_info::lookup_dr instead of
13280 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
13281 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
13282 instead of DR_VECT_AUX.
13284 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13286 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
13288 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
13290 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13291 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
13292 (vect_gen_prolog_loop_niters): Likewise.
13294 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13296 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
13297 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
13298 (vect_known_alignment_in_bytes, vect_dr_behavior)
13299 (vect_get_scalar_dr_size): Take references as dr_vec_infos
13300 instead of data_references. Update calls to other routines for
13301 which the same change has been made.
13302 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
13303 dr_vec_infos instead of stmt_vec_infos.
13304 (vect_analyze_data_ref_dependence): Update call accordingly.
13305 (vect_slp_analyze_data_ref_dependence)
13306 (vect_record_base_alignments): Use DR_VECT_AUX.
13307 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
13308 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
13309 (vector_alignment_reachable_p, vect_get_data_access_cost)
13310 (vect_peeling_supportable, vect_analyze_group_access_1)
13311 (vect_analyze_group_access, vect_analyze_data_ref_access)
13312 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
13313 (vect_compile_time_alias, vect_small_gap_p)
13314 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
13315 (vect_supportable_dr_alignment): Take references as dr_vec_infos
13316 instead of data_references. Update calls to other routines for
13317 which the same change has been made.
13318 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
13319 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
13320 (vect_slp_analyze_and_verify_node_alignment)
13321 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
13322 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
13323 (vect_setup_realignment): Use dr_vec_infos. Update calls after
13325 (_vect_peel_info::dr): Replace with...
13326 (_vect_peel_info::dr_info): ...this new field.
13327 (vect_peeling_hash_get_most_frequent)
13328 (vect_peeling_hash_choose_best_peeling): Update accordingly.
13329 (vect_peeling_hash_get_lowest_cost):
13330 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
13331 routines for which the same change has been made.
13332 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
13334 * tree-vect-loop-manip.c (get_misalign_in_elems)
13335 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
13337 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13338 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
13339 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
13340 (get_group_load_store_type, get_negative_load_store_type)
13341 (vect_get_data_ptr_increment, vectorizable_store)
13342 (vectorizable_load): Likewise.
13343 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
13344 Update calls to other routines for which the same change has been made.
13346 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13348 * tree-vectorizer.h (vec_info::move_dr): New member function.
13349 (dataref_aux): Rename to...
13350 (dr_vec_info): ...this and add "dr" and "stmt" fields.
13351 (_stmt_vec_info::dr_aux): Update accordingly.
13352 (_stmt_vec_info::data_ref_info): Delete.
13353 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
13354 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
13355 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
13356 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
13358 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
13359 (STMT_VINFO_DR_INFO): New macro.
13360 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
13361 (set_dr_misalignment): Update after rename of dataref_aux.
13362 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
13363 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
13364 initialization of STMT_VINFO_DATA_REF.
13365 * tree-vectorizer.c (vec_info::move_dr): New function.
13366 * tree-vect-patterns.c (vect_recog_bool_pattern)
13367 (vect_recog_mask_conversion_pattern)
13368 (vect_recog_gather_scatter_pattern): Use it.
13369 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
13370 the "dr" and "stmt" fields of dr_vec_info instead of
13371 STMT_VINFO_DATA_REF.
13373 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13375 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
13376 (is_pattern_stmt_p): Use it.
13377 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
13378 on pattern statements.
13380 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13382 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
13383 original stmt as a stmt_vec_info rather than a gimple stmt.
13384 (vect_pattern_recog_1): Take the statement directly as a
13385 stmt_vec_info, rather than via a gimple_stmt_iterator.
13386 Update call to vect_mark_pattern_stmts.
13387 (vect_pattern_recog): Update calls accordingly.
13389 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13391 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
13392 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
13393 a vect_def_type for the first argument.
13394 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
13395 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
13396 operand if it isn't defined by a vectorized statement.
13397 (vect_build_gather_load_calls): Remove the mask_dt argument and
13398 update calls to vect_get_vec_def_for_stmt_copy.
13399 (vectorizable_bswap): Likewise the dt argument.
13400 (vectorizable_call): Update calls to vectorizable_bswap and
13401 vect_get_vec_def_for_stmt_copy.
13402 (vectorizable_simd_clone_call, vectorizable_assignment)
13403 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
13404 (vectorizable_comparison): Update calls to
13405 vect_get_vec_def_for_stmt_copy.
13406 (vectorizable_store): Likewise. Remove now-unnecessary calls to
13407 vect_is_simple_use.
13408 (vect_get_loop_based_defs): Remove dt argument and update call
13409 to vect_get_vec_def_for_stmt_copy.
13410 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
13411 and vect_get_vec_def_for_stmt_copy.
13412 (vectorizable_load): Update calls to vect_build_gather_load_calls
13413 and vect_get_vec_def_for_stmt_copy.
13414 * tree-vect-loop.c (vect_create_epilog_for_reduction)
13415 (vectorizable_reduction, vectorizable_live_operation): Update calls
13416 to vect_get_vec_def_for_stmt_copy.
13418 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13420 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
13421 and gimple stmt arguments with a stmt_vec_info.
13422 (vect_record_base_alignments): Update calls accordingly.
13423 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
13424 and gimple stmt arguments with a stmt_vec_info.
13425 (vect_build_slp_tree_1): Remove vinfo argument and update call
13426 to vect_record_max_nunits.
13427 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
13428 and vect_record_max_nunits.
13430 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13432 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
13433 file and take a stmt_vec_info instead of a gimple stmt.
13434 (supportable_widening_operation, vect_finish_replace_stmt)
13435 (vect_finish_stmt_generation, vect_get_store_rhs)
13436 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
13437 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
13438 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
13439 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
13440 (vect_create_data_ref_ptr, bump_vector_ptr)
13441 (vect_permute_store_chain, vect_setup_realignment)
13442 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
13443 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
13444 (vectorizable_reduction, vectorizable_induction)
13445 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
13446 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
13447 than gimple stmts as arguments.
13448 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
13449 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
13450 (can_group_stmts_p, vect_check_gather_scatter)
13451 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
13452 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
13453 (vect_permute_load_chain, vect_shift_permute_load_chain)
13454 (vect_transform_grouped_load)
13455 (vect_record_grouped_load_vectors): Likewise.
13456 * tree-vect-loop.c (vect_fixup_reduc_chain)
13457 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
13458 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
13459 (vectorizable_reduction, vectorizable_induction)
13460 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
13461 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
13462 (vect_get_load_store_mask): Likewise.
13463 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
13464 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
13465 * tree-vect-stmts.c (vect_mark_relevant)
13466 (is_simple_and_all_uses_invariant)
13467 (exist_non_indexing_operands_for_use_p, process_use)
13468 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
13469 (vect_get_vec_def_for_operand, vect_get_vec_defs)
13470 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
13471 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
13472 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
13473 (get_negative_load_store_type, get_load_store_type)
13474 (vect_check_load_store_mask, vect_check_store_rhs)
13475 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
13476 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
13477 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
13478 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
13479 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
13480 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
13481 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
13482 (supportable_widening_operation): Likewise.
13484 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13486 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
13487 a stmt_vec_info instead of a gcall.
13488 (vect_check_gather_scatter): Update call accordingly.
13489 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
13491 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
13492 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
13493 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
13494 instead of a gimple stmt.
13495 (vect_transform_loop): Update calls accordingly.
13496 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
13497 stmt_vec_infos instead of gimple stmts.
13498 (vect_analyze_slp_instance): Update use accordingly.
13499 * tree-vect-stmts.c (read_vector_array, write_vector_array)
13500 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
13501 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
13502 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
13503 (vect_gen_widened_results_half, vect_get_loop_based_defs)
13504 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
13505 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
13506 down to subroutines.
13508 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13510 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
13511 of the worklist from a vector of gimple stmts to a vector of
13513 * tree-vect-stmts.c (vect_mark_relevant, process_use)
13514 (vect_mark_stmts_to_be_vectorized): Likewise
13516 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13518 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
13519 statement before passing it to vect_analyze_stmt.
13520 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
13521 the chain of phi vector definitions. Track the exit phi via its
13523 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
13524 STMT_VINFO_REDUC_DEF.
13525 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
13526 stmt_vec_infos to handle the statement chains.
13527 (vect_get_slp_defs): Record the first statement in the node
13528 using a stmt_vec_info.
13529 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
13530 statements here and pass their stmt_vec_info down to subroutines.
13531 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
13532 down to vect_finish_stmt_generation.
13533 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
13534 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
13535 stmt_vec_infos to subroutines.
13536 (vect_remove_stores): Use stmt_vec_infos to handle the statement
13539 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13541 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
13542 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
13543 (vect_permute_store_chain, vect_setup_realignment)
13544 (vect_permute_load_chain, vect_shift_permute_load_chain)
13545 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
13546 stmts internally, and when passing values to other vectorizer routines.
13547 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
13548 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
13549 (vect_analyze_loop_operations, get_initial_def_for_reduction)
13550 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
13551 (vectorizable_reduction, vectorizable_induction)
13552 (vectorizable_live_operation, vect_transform_loop_stmt)
13553 (vect_transform_loop): Likewise.
13554 * tree-vect-patterns.c (vect_reassociating_reduction_p)
13555 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
13556 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
13557 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
13558 (vect_slp_analyze_node_operations_1): Likewise.
13559 * tree-vect-stmts.c (vect_mark_relevant, process_use)
13560 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
13561 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
13562 (vect_finish_stmt_generation_1, get_group_load_store_type)
13563 (get_load_store_type, vect_build_gather_load_calls)
13564 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
13565 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
13566 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
13567 (vectorizable_store, vectorizable_load, vectorizable_condition)
13568 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
13569 (supportable_widening_operation): Likewise.
13570 (vect_get_vector_types_for_stmt): Likewise.
13571 * tree-vectorizer.h (vect_dr_behavior): Likewise.
13573 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13575 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
13576 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
13577 (vect_permute_store_chain, vect_permute_load_chain)
13578 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
13579 repeated stmt_vec_info lookups.
13580 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
13581 (vect_update_ivs_after_vectorizer): Likewise.
13582 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
13583 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
13584 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
13585 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
13586 (vect_bb_slp_scalar_cost): Likewise.
13587 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
13589 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13591 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
13592 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
13593 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
13594 to get gassigns and gcalls, rather than operating on generc gimple
13596 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
13597 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
13598 (vectorizable_load, vect_analyze_stmt): Likewise.
13599 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
13601 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13603 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
13604 return stmt_vec_infos rather than gimple stmts. Do not accept
13606 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
13608 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
13609 Update use of get_later_stmt.
13610 (vect_get_constant_vectors): Update call accordingly.
13611 (vect_schedule_slp_instance): Likewise
13612 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
13613 (vect_slp_analyze_instance_dependence): Likewise.
13614 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
13616 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13618 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
13619 (stmt_info_for_cost::stmt_info): ...this new field.
13620 (add_stmt_costs): Update accordingly.
13621 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
13622 (vect_get_known_peeling_cost): Likewise.
13623 (vect_estimate_min_profitable_iters): Likewise.
13624 * tree-vect-stmts.c (record_stmt_cost): Likewise.
13626 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13628 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
13629 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
13630 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
13632 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
13634 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13636 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
13637 a gimple stmt to a stmt_vec_info.
13638 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
13640 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13642 * tree-vectorizer.h (vec_info::grouped_stores): Change from
13643 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
13644 (_loop_vec_info::reduction_chains): Likewise.
13645 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
13647 * tree-vect-slp.c (vect_analyze_slp): Likewise.
13649 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13651 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
13652 a gimple stmt to a stmt_vec_info.
13653 (_stmt_vec_info::next_element): Likewise.
13654 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
13655 (vect_slp_analyze_and_verify_node_alignment)
13656 (vect_analyze_group_access_1, vect_analyze_group_access)
13657 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
13658 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
13659 (vect_supportable_dr_alignment): Update accordingly.
13660 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
13661 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
13662 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
13663 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
13664 * tree-vect-slp.c (vect_build_slp_tree_1)
13665 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
13666 (vect_split_slp_store_group, vect_analyze_slp_instance)
13667 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
13668 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
13669 (get_group_load_store_type, get_load_store_type)
13670 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
13671 (vect_transform_stmt, vect_remove_stores): Likewise.
13673 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13675 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
13676 than a gimple stmt.
13677 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
13678 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
13679 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
13680 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
13681 (vector_alignment_reachable_p, vect_get_data_access_cost)
13682 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
13683 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
13684 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
13685 (vect_analyze_group_access_1, vect_analyze_group_access)
13686 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
13687 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
13688 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
13689 result of vect_dr_stmt and use the stmt_vec_info instead of
13690 the associated gimple stmt.
13691 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
13692 (vect_gen_prolog_loop_niters): Likewise.
13693 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13695 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13697 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
13698 to a vec<stmt_vec_info>.
13699 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
13700 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
13701 vec<stmt_vec_info>.
13702 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
13703 to a vec<stmt_vec_info>.
13704 (bst_traits::value_type, bst_traits::value_type): Likewise.
13705 (bst_traits::hash): Update accordingly.
13706 (vect_get_and_check_slp_defs): Change the stmts parameter from
13707 a vec<gimple *> to a vec<stmt_vec_info>.
13708 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
13709 (vect_build_slp_tree): Likewise.
13710 (vect_build_slp_tree_2): Likewise. Update uses of
13711 SLP_TREE_SCALAR_STMTS.
13712 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
13713 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
13714 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
13715 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
13716 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
13717 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
13718 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
13719 (vect_get_constant_vectors, vect_get_slp_defs)
13720 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
13721 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
13722 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
13723 instead of gimple stmts.
13724 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
13725 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
13726 (vect_slp_analyze_instance_dependence): Update uses of
13727 SLP_TREE_SCALAR_STMTS.
13728 (vect_slp_analyze_and_verify_node_alignment): Likewise.
13729 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
13730 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
13731 (get_initial_defs_for_reduction): Likewise.
13732 (vect_create_epilog_for_reduction): Likewise.
13733 (vectorize_fold_left_reduction): Likewise.
13734 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
13735 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
13736 (can_vectorize_live_stmts): Likewise.
13738 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13740 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
13741 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
13742 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
13744 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
13746 (gather_scalar_reductions): Update after above interface changes.
13747 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
13748 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
13750 (vect_force_simple_reduction): Likewise.
13751 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
13752 LOOP_VINFO_REDUCTIONS.
13753 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
13755 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13757 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
13758 a gimple stmt to a stmt_vec_info.
13759 * tree-vect-loop.c (vect_active_double_reduction_p)
13760 (vect_force_simple_reduction, vectorizable_reduction): Update
13763 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13765 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
13766 vec<gimple *> to a vec<stmt_vec_info>.
13767 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
13768 the reduction_phis argument from a vec<gimple *> to a
13769 vec<stmt_vec_info>.
13770 (vectorizable_reduction): Likewise the phis local variable that
13771 is passed to vect_create_epilog_for_reduction. Update for new type
13772 of SLP_TREE_VEC_STMTS.
13773 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
13774 (vectorizable_live_operation): Likewise.
13775 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
13776 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
13778 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13780 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
13781 a gimple stmt to a stmt_vec_info.
13782 (vectorizable_condition, vectorizable_live_operation)
13783 (vectorizable_reduction, vectorizable_induction): Pass back the
13784 vectorized statement as a stmt_vec_info.
13785 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
13786 use of STMT_VINFO_VEC_STMT.
13787 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
13788 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
13789 as stmt_vec_infos rather than gimple stmts.
13790 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
13791 to a stmt_vec_info.
13792 (vectorizable_live_operation): Likewise.
13793 (vectorizable_reduction, vectorizable_induction): Likewise,
13794 updating use of STMT_VINFO_VEC_STMT.
13795 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
13796 of STMT_VINFO_VEC_STMT.
13797 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
13798 (vectorizable_simd_clone_call, vectorizable_conversion)
13799 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
13800 (vectorizable_store, vectorizable_load, vectorizable_condition)
13801 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
13802 from a gimple stmt to a stmt_vec_info.
13803 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
13804 pointer to a stmt_vec_info to the vectorizable_* routines.
13806 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13808 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
13809 a gimple stmt to a stmt_vec_info.
13810 (is_pattern_stmt_p): Update accordingly.
13811 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
13812 (vect_record_grouped_load_vectors): Likewise.
13813 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
13814 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
13815 (vect_model_reduction_cost): Likewise.
13816 (vect_create_epilog_for_reduction): Likewise.
13817 (vectorizable_reduction, vectorizable_induction): Likewise.
13818 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
13819 Return the stmt_vec_info for the pattern statement.
13820 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
13821 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
13822 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
13823 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
13824 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
13825 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
13826 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
13827 (free_stmt_vec_info, vect_is_simple_use): Likewise.
13829 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13831 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
13832 (vect_finish_stmt_generation): Likewise.
13833 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
13834 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
13835 (vect_build_gather_load_calls): Use the return value of the above
13836 functions instead of a separate call to vinfo_for_stmt. Use narrow
13837 scopes for the input gimple stmt and wider scopes for the associated
13838 stmt_vec_info. Use vec_info::lookup_def when setting these
13839 stmt_vec_infos from an SSA_NAME definition.
13840 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
13841 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
13842 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
13843 (vectorizable_store, vectorizable_load, vectorizable_condition)
13844 (vectorizable_comparison): Likewise.
13845 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
13846 (vectorizable_reduction): Likewise.
13848 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13850 * tree-vectorizer.h (vect_is_simple_use): Add an optional
13851 stmt_vec_info * parameter before the optional gimple **.
13852 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
13853 (process_use, vect_get_vec_def_for_operand_1): Update callers.
13854 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
13855 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13856 (vectorizable_live_operation): Likewise.
13857 * tree-vect-patterns.c (type_conversion_p): Likewise.
13858 (vect_look_through_possible_promotion): Likewise.
13859 (vect_recog_rotate_pattern): Likewise.
13860 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
13862 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13864 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
13865 a typedef to a wrapper class.
13866 (NULL_STMT_VEC_INFO): New macro.
13867 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
13868 (stmt_vec_info::operator*): New function.
13869 (stmt_vec_info::operator gimple *): Likewise.
13870 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
13871 (add_stmt_costs): Likewise.
13872 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
13873 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
13874 (vect_get_known_peeling_cost): Likewise.
13875 (vect_estimate_min_profitable_iters): Likewise.
13876 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
13877 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
13878 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
13879 (vectorizable_store, free_stmt_vec_infos): Likewise.
13880 (new_stmt_vec_info): Change return type of xcalloc to
13883 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13885 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
13886 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
13887 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
13888 a single_imm_use-based sequence.
13889 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
13891 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13893 * tree-vectorizer.h (vec_info::lookup_def): Declare.
13894 * tree-vectorizer.c (vec_info::lookup_def): New function.
13895 * tree-vect-patterns.c (vect_get_internal_def): Use it.
13896 (vect_widened_op_tree): Likewise.
13897 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
13898 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
13899 (vectorizable_reduction): Likewise.
13900 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
13902 (vect_is_slp_reduction): Update calls accordingly. Use
13903 vec_info::lookup_def.
13904 (vect_is_simple_reduction): Likewise
13905 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
13907 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13909 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
13910 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
13911 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
13913 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
13914 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
13915 (vect_update_vf_for_slp, vect_analyze_loop_operations)
13916 (vect_is_slp_reduction, vectorizable_induction)
13917 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
13918 * tree-vect-patterns.c (vect_init_pattern_stmt):
13919 (vect_determine_min_output_precision_1, vect_determine_precisions)
13920 (vect_pattern_recog): Likewise.
13921 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
13922 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
13923 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
13924 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
13925 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
13926 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
13927 info field from a loop to a loop_vec_info.
13929 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13931 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
13932 (vec_info::add_stmt): Declare.
13933 * tree-vectorizer.c (vec_info::add_stmt): New function.
13934 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
13935 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
13936 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
13937 (vectorizable_induction): Likewise.
13938 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
13939 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
13940 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
13941 (vectorizable_load): Likewise.
13942 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
13943 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
13944 (vect_recog_gather_scatter_pattern): Likewise.
13945 (append_pattern_def_seq): Likewise. Remove a check that is
13946 performed by add_stmt itself.
13948 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13950 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
13951 which make_ssa_name was called with new_stmt before new_stmt
13954 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13956 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
13958 (vect_is_slp_reduction): ...here...
13959 (vect_is_simple_reduction): ...and here. Remove repetition of tests
13960 that are already known to be false.
13962 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13964 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
13965 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
13966 STMT_VINFO_NUM_SLP_USES when it's true.
13967 (vect_free_slp_instance): Add a final_p parameter and pass it to
13968 vect_free_slp_tree.
13969 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
13970 (vect_analyze_slp_instance): Likewise.
13971 (vect_slp_analyze_operations): Likewise.
13972 (vect_slp_analyze_bb_1): Likewise.
13973 * tree-vectorizer.c (vec_info): Likewise.
13974 * tree-vect-loop.c (vect_transform_loop): Likewise.
13976 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13978 * tree-vect-loop.c (vectorizable_reduction): Assert that the
13979 function is not called for second and subsequent members of
13982 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13984 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
13985 cases for nested loops from here to ...
13986 (vect_create_epilog_for_reduction): ...here. Only call
13987 vect_is_simple_use for inner-loop reductions.
13989 2018-07-31 Martin Liska <mliska@suse.cz>
13991 PR gcov-profile/85338
13992 PR gcov-profile/85350
13993 PR gcov-profile/85372
13994 * profile.c (struct location_triplet): New.
13995 (struct location_triplet_hash): Likewise.
13996 (output_location): Do not output a BB that
13997 is already recorded for a line.
13998 (branch_prob): Use streamed_locations.
14000 2018-07-31 Martin Liska <mliska@suse.cz>
14002 PR gcov-profile/85370
14003 * coverage.c (coverage_begin_function): Do not mark target
14004 clones as artificial functions.
14006 2018-07-31 Martin Liska <mliska@suse.cz>
14008 PR gcov-profile/83813
14009 PR gcov-profile/84758
14010 PR gcov-profile/85217
14011 PR gcov-profile/85332
14012 * profile.c (branch_prob): Do not record GOTO expressions
14013 for GIMPLE statements which locations are already streamed.
14015 2018-07-31 Olivier Hainque <hainque@adacore.com>
14017 * gcc.c (handle_spec_function): Accept a soft_matched_part
14018 argument, as do_spec_1. Pass it down to ...
14019 (eval_spec_function): Accept a soft_matched_part argument,
14020 and pass it down to ...
14021 (do_spec_2): Accept a soft_matched_part argument, and pass
14022 it down to do_spec_1.
14023 (do_spec_1): Pass soft_matched_part to handle_spec_function.
14024 (handle_braces): Update call to handle_spec_function.
14025 (driver::set_up_specs): Update calls to do_spec_2.
14026 (compare_debug_dump_opt_spec_function): Likewise.
14027 (compare_debug_self_opt_spec_function): Likewise.
14029 2018-07-31 Olivier Hainque <hainque@adacore.com>
14031 * common.opt (nolibc): New option.
14032 * doc/invoke.texi (Link Options): Document it.
14033 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
14034 * config/alpha/linux.h: Likewise.
14035 * config/arc/elf.h: Likewise.
14036 * config/arm/uclinux-elf.h: Likewise.
14037 * config/arm/unknown-elf.h: Likewise.
14038 * config/avr/avrlibc.h: Likewise.
14039 * config/bfin/bfin.h: Likewise.
14040 * config/bfin/linux.h: Likewise.
14041 * config/bfin/uclinux.h: Likewise.
14042 * config/darwin.h: Likewise.
14043 * config/darwin10.h: Likewise.
14044 * config/darwin12.h: Likewise.
14045 * config/gnu-user.h: Likewise.
14046 * config/lm32/uclinux-elf.h: Likewise.
14047 * config/pa/pa-hpux11.h: Likewise.
14048 * config/pa/pa64-hpux.h: Likewise.
14049 * config/sparc/sparc.h: Likewise.
14051 2018-07-31 Olivier Hainque <hainque@adacore.com>
14053 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
14054 undefined variables.
14056 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
14059 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
14060 instead of GEN_INT.
14062 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
14064 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
14065 terminated string literal.
14067 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
14069 PR rtl-optimization/85160
14070 * combine.c (is_just_move): New function.
14071 (try_combine): Allow combining two instructions into two if neither of
14072 the original instructions was a move.
14074 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
14077 * doc/extend.texi (Global Register Variables): Discourage use of type
14079 (Local Register Variables): Likewise.
14081 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
14083 PR tree-optimization/86506
14084 * hwint.h (ceil_log2): Resync with hwint.c implementation.
14086 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
14089 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
14090 hard_regno, make sure no insn between `from` and `to` clobbers it.
14092 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
14093 Tom de Vries <tdevries@suse.de>
14095 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
14096 (PTX_DEFAULT_RUNTIME_DIM): ... this.
14097 (nvptx_goacc_validate_dims): Set default worker and gang dims to
14098 PTX_DEFAULT_RUNTIME_DIM.
14099 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
14101 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
14103 * config/pa/pa.c (pa_output_addr_vec): Align address table.
14104 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
14105 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
14107 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
14109 * config/rs6000/constraints.md (wG constraint): Delete, no longer
14111 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
14112 predicate to reflect toc fusion has been deleted.
14113 (toc_fusion_mem_raw): Delete, no longer used.
14114 (toc_fusion_mem_wrapped): Likewise.
14115 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
14117 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
14118 Delete, no longer used.
14119 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
14120 meant to be used for toc fusion.
14121 (rs6000_debug_print_mode): Delete toc fusion debugging.
14122 (rs6000_debug_reg_global): Likewise.
14123 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
14124 fusion and secondary reload support that were never used.
14125 (rs6000_option_override_internal): Delete TOC fusion, that was only
14126 partially defined, and it did not work unless you also used the
14128 (rs6000_legitimate_address_p): Delete TOC fusion support.
14129 (rs6000_opt_masks): Likewise.
14130 (fusion_wrap_memory_address): Delete function, no longer used.
14131 (fusion_split_address); Delete TOC fusion support.
14132 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
14133 longer used with toc fusion being deleted.
14134 (TARGET_TOC_FUSION_FP): Likewise.
14135 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
14137 (toc fusion spliter): Delete TOC fusion support.
14138 (toc_fusionload_<mode>): Likewise.
14139 (toc_fusionload_di): Likewise.
14140 (fusion_gpr_load_<mode>): Delete generator function, this insn no
14141 longer needs to be named. Rename predicate to delete TOC fusion.
14142 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
14143 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
14144 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
14145 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
14146 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
14148 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
14150 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
14151 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
14152 __int128 in built-in function prototypes.
14153 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
14154 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
14156 2018-07-27 Martin Sebor <msebor@redhat.com>
14158 PR tree-optimization/86696
14159 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
14160 types, including enums.
14161 (handle_char_store): Be prepared for the above function to fail.
14163 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
14165 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
14166 when optimization level is lower than 2 or optimize for size.
14168 2018-07-26 Martin Sebor <msebor@redhat.com>
14170 PR tree-optimization/86043
14171 PR tree-optimization/86042
14172 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
14173 (get_string_cst_length): Rename...
14174 (get_min_string_length): ...to this. Add argument.
14175 (handle_char_store): Extend to handle multi-character stores by
14177 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
14178 * tree.h (initializer_zerop): Add argument.
14180 2018-07-26 Jakub Jelinek <jakub@redhat.com>
14182 PR middle-end/86660
14183 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
14184 declare target to variables if they have always,{to,from,tofrom} map
14187 2018-07-26 Martin Liska <mliska@suse.cz>
14190 * lto-wrapper.c: Add linker_output as prefix
14191 for ltrans_output_file.
14193 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
14195 PR rtl-optimization/85805
14196 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
14197 value for hard registers if that was written in the same mode.
14199 2018-07-26 Martin Liska <mliska@suse.cz>
14201 PR gcov-profile/86536
14202 * gcov.c (format_gcov): Use printf format %.*f directly
14203 and do not handle special values.
14205 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
14207 * common/config/arc/arc-common.c (arc_option_optimization_table):
14208 Update default optimizations for size.
14210 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
14212 * config/arc/arc.md (movsf_insn): Add short instruction selection.
14213 * config/arc/constraints.md (CfZ): New constraint.
14214 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
14215 (subsf3_fpu): Likewise.
14216 (cmpsf_fpu): Likewise.
14217 (cmpsf_fpu_uneq): Likewise.
14219 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
14221 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
14223 (arc_is_uncached_mem_p): uncached applies to both the variable and
14226 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
14228 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
14231 2018-07-25 David Malcolm <dmalcolm@redhat.com>
14233 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
14234 field "m_scopes" from vec to auto_vec.
14236 2018-07-25 Martin Liska <mliska@suse.cz>
14238 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
14241 2018-07-25 Richard Biener <rguenther@suse.de>
14244 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
14245 special wrt context_die late.
14246 (gen_subprogram_die): Re-use DIEs in local scope.
14248 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
14250 PR tree-optimization/86644
14251 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
14253 2018-07-25 Martin Liska <mliska@suse.cz>
14255 PR middle-end/86645
14256 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
14257 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
14259 2018-07-25 Martin Liska <mliska@suse.cz>
14262 * params.def: Explain ASan abbreviation and provide
14263 a documentation link.
14265 2018-07-24 Martin Sebor <msebor@redhat.com>
14267 PR tree-optimization/86622
14268 PR tree-optimization/86532
14269 * builtins.h (string_length): Declare.
14270 * builtins.c (c_strlen): Correct handling of non-constant offsets.
14271 (check_access): Be prepared for non-constant length ranges.
14272 (string_length): Make extern.
14273 * expr.c (string_constant): Only handle the minor non-constant
14274 array index. Use string_constant to compute the length of
14275 a generic string constant.
14277 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
14279 PR tree-optimization/86618
14280 * tree-vect-stmts.c (vectorizable_call): Don't take the address
14281 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
14283 2018-07-24 David Malcolm <dmalcolm@redhat.com>
14285 PR tree-optimization/86636
14286 * json.cc (json::object::set): Fix comment. Add assertions.
14287 (json::array::append): Move here from json.h. Add comment and an
14289 (json::string::string): Likewise.
14290 * json.h (json::array::append): Move to json.cc.
14291 (json::string::string): Likewise.
14292 * optinfo-emit-json.cc
14293 (optrecord_json_writer::impl_location_to_json): Assert that we
14294 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
14295 wrapper around it. Expand the location once, rather than three
14297 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
14298 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
14300 (optrecord_json_writer::optinfo_to_json): Likewise, in four
14301 places. Fix some overlong lines.
14303 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
14305 * config/aarch64/aarch64-simd.md
14306 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
14307 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
14308 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
14309 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
14310 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
14311 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
14312 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
14313 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
14314 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
14316 2018-07-24 Jakub Jelinek <jakub@redhat.com>
14318 PR middle-end/86627
14319 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
14320 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
14321 and abs_d == d, do the power of two handling if profitable.
14323 2018-07-24 Richard Biener <rguenther@suse.de>
14325 * match.pd: Add BIT_FIELD_REF canonicalizations.
14327 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14330 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
14332 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14334 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
14335 terminated STRING_CST object.
14337 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14339 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
14341 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
14343 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
14344 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
14345 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
14346 the elements into a register.
14347 (rs6000_split_v4si_init_di_reg): Delete.
14348 (rs6000_split_v4si_init): Delete.
14349 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
14350 (vsx_init_v4si): Rewrite as a define_expand.
14352 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
14354 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
14355 zero_extend argument from memory): New.
14357 2018-07-22 Martin Sebor <msebor@redhat.com>
14360 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
14361 diagnosing calls with unknown arguments unless -Walloca-larger-than
14362 is restricted to less than PTRDIFF_MAX bytes.
14364 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
14366 * doc/gcov.texi (Invoking Gcov): Editorial changes.
14368 2018-07-20 David Malcolm <dmalcolm@redhat.com>
14370 * pretty-print.c (text_info::set_location): Remove redundant
14371 "line_table" parameter from call to rich_location::set_range.
14373 2018-07-20 Martin Sebor <msebor@redhat.com>
14375 PR middle-end/82063
14376 * builtins.c (expand_builtin_alloca): Adjust.
14377 * calls.c (alloc_max_size): Simplify.
14378 * cgraphunit.c (cgraph_node::expand): Adjust.
14379 * common.opt (larger_than_size, warn_frame_larger_than): Remove
14381 (frame_larger_than_size): Same.
14382 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
14383 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
14385 * doc/invoke.texi (GCC Command Options): Document option arguments.
14386 Explain byte-size arguments and suffixes.
14387 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
14388 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
14389 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
14390 * doc/options.texi (UInteger): Expand.
14391 (Host_Wide_Int, ByteSize): Document new properties.
14392 * final.c (final_start_function_1): Include sizes in an error message.
14393 * function.c (frame_offset_overflow): Same.
14394 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
14395 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
14396 Diagnose unbounded alloca calls only for limits of less than
14398 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
14399 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
14401 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
14402 only for limits of less than PTRDIFF_MAX.
14403 * langhooks-def.h (lhd_handle_option): Change function argument
14405 * langhooks.c (lhd_handle_option): Same.
14406 * langhooks.h (handle_option): Same.
14407 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
14409 (var_type, var_type_struct): Same.
14410 (var_set): Handle ByteSize flag.
14411 * optc-gen.awk: Add comments to output to ease debugging. Make
14412 use of HOST_WIDE_INT where appropriate.
14413 * opts-gen-save.awk: Use %lx to format unsigned long.
14414 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
14415 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
14416 arguments. Parse bytes-size suffixes.
14417 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
14418 (enum_value_to_arg): Same.
14419 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
14420 (handle_option): Adjust.
14421 (generate_option): Change function argument to HOST_WIDE_INT.
14422 (cmdline_handle_error): Adjust.
14423 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
14424 (set_option): Change function argument to HOST_WIDE_INT.
14425 (option_enabled): Handle cl_host_wide_int.
14426 (get_option_state): Handle CLVC_SIZE.
14427 (control_warning_option): Same.
14428 * opts.c (common_handle_option): Change function argument to
14429 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
14430 OPT_Wvla_larger_than_.
14431 * opts.h (enum cl_var_type): Add an enumerator.
14432 * stor-layout.c (layout_decl): Print a more meaningful warning.
14433 * toplev.c (output_stack_usage): Adjust.
14435 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
14437 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
14438 call to inline_expand_builtin_string_cmp.
14439 (expand_builtin_strcmp): Likewise.
14440 (expand_builtin_strncmp): Likewise.
14441 (inline_string_cmp): Delete the last parameter, change char_type_node
14442 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
14444 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
14445 the inlining expansion on target where the type of the call has same or
14446 narrower precision than unsigned char.
14448 2018-07-20 David Malcolm <dmalcolm@redhat.com>
14450 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
14451 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
14452 * common.opt (fsave-optimization-record): New option.
14453 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
14454 * doc/invoke.texi (-fsave-optimization-record): New option.
14455 * dumpfile.c: Include "optinfo-emit-json.h".
14456 (struct kv_pair): Move to coretypes.h.
14457 (optgroup_options): Make non-static.
14458 (dump_context::end_scope): Call
14459 optimization_records_maybe_pop_dump_scope.
14460 * dumpfile.h (optgroup_options): New decl.
14461 * json.cc: New file.
14462 * json.h: New file.
14463 * optinfo-emit-json.cc: New file.
14464 * optinfo-emit-json.h: New file.
14465 * optinfo.cc: Include "optinfo-emit-json.h".
14466 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
14467 (optinfo_enabled_p): Check optimization_records_enabled_p.
14468 (optinfo_wants_inlining_info_p): Likewise.
14469 * optinfo.h: Update comment.
14470 * profile-count.c (profile_quality_as_string): New function.
14471 * profile-count.h (profile_quality_as_string): New decl.
14472 (profile_count::quality): New accessor.
14473 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
14474 and optinfo_emit_json_cc_tests.
14475 * selftest.h (selftest::json_cc_tests): New decl.
14476 (selftest::optinfo_emit_json_cc_tests): New decl.
14477 * toplev.c: Include "optinfo-emit-json.h".
14478 (compile_file): Call optimization_records_finish.
14479 (do_compile): Call optimization_records_start.
14480 * tree-ssa-live.c: Include optinfo.h.
14481 (remove_unused_scope_block_p): Retain inlining information if
14482 optinfo_wants_inlining_info_p returns true.
14484 2018-07-20 Richard Biener <rguenther@suse.de>
14487 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
14488 to cover -flto-partition=none.
14490 2018-07-20 Martin Liska <mliska@suse.cz>
14492 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
14493 (get_decl_source_range): Remove unused function.
14495 2018-07-20 Richard Biener <rguenther@suse.de>
14497 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
14498 (struct vn_phi_s): Likewise.
14499 (struct vn_reference_s): Likewise.
14500 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
14501 for searching the slot of an entry known to be in the hash itself.
14502 (vn_phi_hasher::equal): Likewise.
14503 (vn_reference_hasher::equal): Likewise.
14504 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
14506 (optimistic_info, current_info): Remove, keeping only valid_info.
14507 (vn_reference_lookup_1): Remove fallback lookup.
14508 (vn_reference_lookup_2): Likewise.
14509 (vn_nary_op_lookup_1): Likewise.
14510 (vn_phi_lookup): Likewise.
14511 (vn_nary_build_or_lookup_1): Make sure to not chain the built
14513 (vn_reference_insert): Adjust, chain the inserted hash element
14514 at last_inserted_ref.
14515 (vn_reference_insert_pieces): Likewise.
14516 (visit_reference_op_call): Likewise.
14517 (vn_nary_op_insert_into): Chain the inserted hash element at
14518 last_inserted_nary.
14519 (vn_nary_op_insert_pieces): Adjust.
14520 (vn_nary_op_insert): Likewise.
14521 (vn_nary_op_insert_stmt): Likewise.
14522 (vn_phi_insert): Adjust, chain the inserted hash element at
14524 (process_scc): Remove clearing and copying the optimistic
14525 table. Instead remove elements inserted during an optimistic
14526 iteration from the single table we maintain.
14527 (init_scc_vn): Adjust.
14528 (free_scc_vn): Likewise.
14529 (sccvn_dom_walker::record_cond): Likewise.
14530 (sccvn_dom_walker::after_dom_children): Likewise.
14532 2018-07-19 Martin Sebor <msebor@redhat.com>
14534 PR tree-optimization/84047
14535 PR tree-optimization/83776
14536 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
14537 (check_array_bounds): Call it.
14539 2018-07-19 Martin Sebor <msebor@redhat.com>
14541 * align.h (align_flags): Use member initialization.
14543 2018-07-19 David Malcolm <dmalcolm@redhat.com>
14545 * Makefile.in (OBJS): Add optinfo.o.
14546 * coretypes.h (class symtab_node): New forward decl.
14547 (struct cgraph_node): New forward decl.
14548 (class varpool_node): New forward decl.
14549 * dump-context.h: New file.
14550 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
14552 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
14553 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
14554 (set_alt_dump_file): Likewise.
14555 (dump_context::~dump_context): New dtor.
14556 (dump_gimple_stmt): Move implementation to...
14557 (dump_context::dump_gimple_stmt): ...this new member function.
14558 Add the stmt to any pending optinfo, creating one if need be.
14559 (dump_gimple_stmt_loc): Move implementation to...
14560 (dump_context::dump_gimple_stmt_loc): ...this new member function.
14561 Start a new optinfo and add the stmt to it.
14562 (dump_gimple_expr): Move implementation to...
14563 (dump_context::dump_gimple_expr): ...this new member function.
14564 Add the stmt to any pending optinfo, creating one if need be.
14565 (dump_gimple_expr_loc): Move implementation to...
14566 (dump_context::dump_gimple_expr_loc): ...this new member function.
14567 Start a new optinfo and add the stmt to it.
14568 (dump_generic_expr): Move implementation to...
14569 (dump_context::dump_generic_expr): ...this new member function.
14570 Add the tree to any pending optinfo, creating one if need be.
14571 (dump_generic_expr_loc): Move implementation to...
14572 (dump_context::dump_generic_expr_loc): ...this new member
14573 function. Add the tree to any pending optinfo, creating one if
14575 (dump_printf): Move implementation to...
14576 (dump_context::dump_printf_va): ...this new member function. Add
14577 the text to any pending optinfo, creating one if need be.
14578 (dump_printf_loc): Move implementation to...
14579 (dump_context::dump_printf_loc_va): ...this new member function.
14580 Start a new optinfo and add the stmt to it.
14581 (dump_dec): Move implementation to...
14582 (dump_context::dump_dec): ...this new member function. Add the
14583 value to any pending optinfo, creating one if need be.
14584 (dump_context::dump_symtab_node): New member function.
14585 (dump_context::get_scope_depth): New member function.
14586 (dump_context::begin_scope): New member function.
14587 (dump_context::end_scope): New member function.
14588 (dump_context::ensure_pending_optinfo): New member function.
14589 (dump_context::begin_next_optinfo): New member function.
14590 (dump_context::end_any_optinfo): New member function.
14591 (dump_context::s_current): New global.
14592 (dump_context::s_default): New global.
14593 (dump_scope_depth): Delete global.
14594 (dumpfile_ensure_any_optinfo_are_flushed): New function.
14595 (dump_symtab_node): New function.
14596 (get_dump_scope_depth): Reimplement in terms of dump_context.
14597 (dump_begin_scope): Likewise.
14598 (dump_end_scope): Likewise.
14599 (selftest::temp_dump_context::temp_dump_context): New ctor.
14600 (selftest::temp_dump_context::~temp_dump_context): New dtor.
14601 (selftest::verify_item): New function.
14602 (ASSERT_IS_TEXT): New macro.
14603 (ASSERT_IS_TREE): New macro.
14604 (ASSERT_IS_GIMPLE): New macro.
14605 (selftest::test_capture_of_dump_calls): New test.
14606 (selftest::dumpfile_c_tests): Call it.
14607 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
14608 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
14609 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
14610 descriptive comment.
14611 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
14612 (dump_node, dump_bb): Move these unrelated decls.
14613 (class dump_manager): Add leading comment.
14614 * optinfo.cc: New file.
14615 * optinfo.h: New file.
14617 2018-07-19 Michael Collison <michael.collison@arm.com>
14618 Richard Henderson <rth@redhat.com>
14620 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
14621 (subti): Handle op1 zero.
14622 (subvti4, usub4ti4): New.
14623 (*sub<GPI>3_compare1_imm): New.
14624 (sub<GPI>3_carryinCV): New.
14625 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
14626 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
14628 2018-07-19 Michael Collison <michael.collison@arm.com>
14629 Richard Henderson <rth@redhat.com>
14631 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
14632 (addti3): Create simpler code if low part is already known to be 0.
14633 (addvti4, uaddvti4): New.
14634 (*add<GPI>3_compareC_cconly_imm): New.
14635 (*add<GPI>3_compareC_cconly): New.
14636 (*add<GPI>3_compareC_imm): New.
14637 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
14638 handle constants within this pattern..
14639 (*add<GPI>3_compareV_cconly_imm): New.
14640 (*add<GPI>3_compareV_cconly): New.
14641 (*add<GPI>3_compareV_imm): New.
14642 (add<GPI>3_compareV): New.
14643 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
14644 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
14645 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
14646 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
14647 with 'comparison' operator.
14648 (*add<GPI>3_compareV_cconly_imm): Ditto.
14649 (*add<GPI>3_compareV_cconly): Ditto.
14650 (*add<GPI>3_compareV_imm): Ditto.
14651 (add<GPI>3_compareV): Ditto.
14652 (add<mode>3_carryinC): Ditto.
14653 (*add<mode>3_carryinC_zero): Ditto.
14654 (*add<mode>3_carryinC): Ditto.
14655 (add<mode>3_carryinV): Ditto.
14656 (*add<mode>3_carryinV_zero): Ditto.
14657 (*add<mode>3_carryinV): Ditto.
14659 2018-07-19 Michael Collison <michael.collison@arm.com>
14660 Richard Henderson <rth@redhat.com>
14662 * config/aarch64/aarch64-modes.def (CC_V): New.
14663 * config/aarch64/aarch64-protos.h
14664 (aarch64_addti_scratch_regs): Declare
14665 (aarch64_subvti_scratch_regs): Declare.
14666 (aarch64_expand_subvti): Declare.
14667 (aarch64_gen_unlikely_cbranch): Declare
14668 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
14669 for signed overflow using CC_Vmode.
14670 (aarch64_get_condition_code_1): Handle CC_Vmode.
14671 (aarch64_gen_unlikely_cbranch): New function.
14672 (aarch64_addti_scratch_regs): New function.
14673 (aarch64_subvti_scratch_regs): New function.
14674 (aarch64_expand_subvti): New function.
14676 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
14678 * config/aarch64/aarch64-option-extensions.def: New entry for profile
14680 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
14681 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
14684 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
14687 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
14688 address check not strict.
14690 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
14692 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
14694 (store_pair_lanes<mode>): Likewise.
14695 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
14696 enum value 'ADDR_QUERY_LDP_STP_N'.
14697 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
14698 (aarch64_print_address_internal): Add declaration.
14699 (aarch64_print_ldpstp_address): Remove.
14700 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
14701 (aarch64_print_operand): Change printing of 'y'.
14702 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
14703 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
14704 'true' rather than '1'.
14705 * config/aarch64/constraints.md (Uml): Likewise.
14706 (Uml): Rename to Umn.
14709 2018-07-19 Richard Biener <rguenther@suse.de>
14711 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
14713 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
14714 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
14715 (vn_reference_hasher): Likewise.
14716 (struct vn_tables_s): Remove obstack and alloc-pool members.
14717 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
14718 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
14719 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
14720 (vn_reference_insert_pieces): Likewise.
14721 (alloc_vn_nary_op_noinit): Adjust.
14722 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
14723 (vn_phi_eq): Adjust.
14724 (shared_lookup_phiargs): Remove.
14725 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
14726 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
14727 (visit_reference_op_call): Likewise.
14728 (copy_nary, copy_phi, copy_reference): Remove.
14729 (process_scc): Rewind the obstack when iterating. Do not
14730 copy the elements to valid_info but just move them from one
14731 hashtable to the other.
14732 (allocate_vn_table): Adjust.
14733 (free_vn_table): Likewise.
14734 (init_scc_vn): Likewise.
14735 (free_scc_vn): Likewise.
14737 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
14740 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
14741 indirect_return as function type attribute.
14742 (ix86_attribute_table): Change indirect_return to function
14744 * doc/extend.texi: Update indirect_return attribute.
14746 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
14748 * wide-int.h (widest2_int): New.
14749 * gimple-fold.c (arith_overflowed_p): Use it.
14750 * tree.h (widest2_int_cst): New.
14751 * tree-vrp.c (wide_int_binop_overflow): Rename from
14752 vrp_int_const_binop.
14753 Rewrite to work on trees.
14754 (extract_range_from_multiplicative_op_1): Abstract code to...
14755 (wide_int_range_min_max): ...here.
14756 (wide_int_range_cross_product): ...and here.
14757 (extract_range_from_binary_expr_1): Abstract overflow code to...
14758 (wide_int_range_mult_wrapping): ...here.
14759 * tree-vrp.h (wide_int_range_cross_product): New.
14760 (wide_int_range_mult_wrapping): New.
14762 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
14763 Julia Koval <julia.koval@intel.com>
14765 * config/i386/x86-tune-costs.h (skylake_memcpy,
14766 skylake_memset): Replace rep_prefix with unrolling for size 512.
14768 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
14770 PR middle-end/86544
14771 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
14772 comparision with EQ_EXPR in last stmt.
14774 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
14776 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
14777 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
14778 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
14779 previously known as "PowerPC AltiVec Built-in Functions". Move
14780 some material to new subsubsections "PowerPC AltiVec Built-in
14781 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
14783 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
14784 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
14785 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
14786 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
14788 2018-07-18 Richard Biener <rguenther@suse.de>
14790 PR tree-optimization/86557
14791 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
14794 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
14796 * config/s390/s390.c (s390_function_profiler): Generate CFI.
14798 2018-07-17 Jeff Law <law@redhat.com>
14800 * config/arm/arm.c (get_label_padding): Update for recent
14801 changes to label_to_alignment.
14803 PR tree-optimization/86010
14804 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
14806 * config/mips/mips.c (vr4130_align_insns): Update for recent
14807 changes to label_to_alignment.
14809 * config/frv/frv.c (frv_label_align): Update for recent changes
14810 to label_to_alignment.
14812 * config/nios2/nios2.c (nios2_label_align): Update for recent
14813 changes which dropped ALIGN_LABELS_LOG.
14815 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
14817 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
14818 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
14820 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14822 * config/arc/arc.c (arc_label_align): Use align_labels instead of
14823 deprecated align_labels_log.
14825 2018-07-17 Richard Biener <rguenther@suse.de>
14828 * dwarf2out.c (init_sections_and_labels): Always generate
14829 a debug_line_str_section for early LTO debug.
14830 (dwarf2out_finish): Reset debug_line_str_hash output early.
14831 Bump counter for extra dwarf5 .debug_loc labels to not conflict
14833 (dwarf2out_early_finish): Output debug_line_str.
14835 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
14837 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
14838 index register on z196 or later.
14840 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
14842 * config/s390/s390.c (s390_default_align): Set default function
14844 (s390_override_options_after_change): Call s390_default align.
14845 (s390_option_override_internal): Call s390_default_align.
14846 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
14848 2018-07-17 Jakub Jelinek <jakub@redhat.com>
14850 PR middle-end/86542
14851 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
14852 to _looptemp_ clauses, other than the first two.
14854 2018-07-17 Martin Liska <mliska@suse.cz>
14856 * opts.c: Do not enable OPT_falign_* for -Os.
14858 2018-07-17 Martin Liska <mliska@suse.cz>
14860 * align.h (MAX_CODE_ALIGN): New.
14861 (MAX_CODE_ALIGN_VALUE): New.
14862 * common/config/i386/i386-common.c (ix86_handle_option):
14863 (MAX_CODE_ALIGN): Moved to align.h.
14864 * final.c (MAX_CODE_ALIGN): Likewise.
14865 * opts.c (parse_and_check_align_values):
14866 (MAX_CODE_ALIGN): Likewise.
14867 (MAX_CODE_ALIGN_VALUE): Likewise.
14869 2018-07-17 Martin Liska <mliska@suse.cz>
14871 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
14872 in order to fulfil coding style.
14873 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
14874 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14875 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
14876 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
14877 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
14878 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
14879 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
14880 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
14881 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14883 2018-07-17 Martin Liska <mliska@suse.cz>
14885 * align.h: New file.
14886 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
14888 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
14889 align_flags of label_to_alignment.
14890 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
14892 * config/m68k/m68k.c: Do not use removed align_labels_value and
14894 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
14895 (LOOP_ALIGN): Likewise.
14896 (LABEL_ALIGN): Likewise.
14897 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
14898 Remove not used macro.
14899 (rs6000_loop_align): Change return type to align_flags.
14900 (rs6000_loop_align_max_skip): Remove.
14901 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
14902 Change return type to align_flags.
14903 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
14904 Remove not used macro.
14905 (rs6000_loop_align): Change return type to align_flags.
14906 (rs6000_loop_align_max_skip): Remove.
14907 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
14908 * config/rx/rx-protos.h (rx_align_for_label): Make it
14910 * config/rx/rx.c (rx_align_for_label): Change return type
14912 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
14914 into align_flags class.
14915 (LABEL_ALIGN): Likewise.
14916 (LOOP_ALIGN): Likewise.
14917 * config/s390/s390.c (s390_label_align): Use align_flags
14919 (s390_asm_output_function_label): Likewise.
14920 * config/sh/sh.c (sh_override_options_after_change):
14921 Use align_flags class directly without macros.
14922 (find_barrier): Likewise.
14923 (barrier_align): Likewise.
14924 (sh_loop_align): Likewise.
14925 * config/spu/spu.c (spu_option_override):
14926 Use align_flags_tuple::get_value instead of removed macros.
14927 (spu_sched_init): Likewise.
14928 * config/spu/spu.h (GTY): Likewise.
14929 * config/visium/visium.c (visium_option_override):
14930 Set "8" as default secondary alignment.
14931 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
14932 in order to guarantee secondary alignment of 8.
14933 * coretypes.h: Include align.h header file.
14934 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
14935 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
14936 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
14937 * doc/tm.texi.in: Likewise.
14938 * final.c (struct label_alignment): Remove not used structure.
14939 (LABEL_ALIGN): Change type to align_flags.
14940 (LOOP_ALIGN): Likewise.
14941 (JUMP_ALIGN): Likewise.
14942 (default_loop_align_max_skip): Remove.
14943 (default_label_align_max_skip): Likewise.
14944 (default_jump_align_max_skip): Likewise.
14945 (default_label_align_after_barrier_max_skip):
14946 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
14947 (LABEL_TO_MAX_SKIP): Remove.
14948 (label_to_alignment): Return align_flags type instead of integer.
14949 (label_to_max_skip): Remove.
14950 (align_fuzz): Use align_flags type.
14951 (compute_alignments): Use align_flags type and use align_flags::max
14952 to combine multiple alignments.
14953 (grow_label_align): Grow vec instead of C array.
14954 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
14955 (shorten_branches): Use align_flags type and use align_flags::max
14956 to combine multiple alignments.
14957 (final_scan_insn_1): Remove usage of secondary alignment that comes
14958 from label alignment, but instead use proper secondary alignment
14959 which is computed in grow_label_align.
14960 * flags.h (struct align_flags_tuple): Move to align.h.
14961 (struct align_flags): Likewise.
14962 (state_align_loops): Rename to align_loops.
14963 (state_align_jumps): Rename to align_jumps.
14964 (state_align_labels): Rename to align_labels.
14965 (state_align_functions): Rename to align_functions.
14966 (align_loops_log): Remove.
14967 (align_jumps_log): Remove.
14968 (align_labels_log): Remove.
14969 (align_functions_log): Remove.
14970 (align_loops_max_skip): Remove.
14971 (align_jumps_max_skip): Remove.
14972 (align_labels_max_skip): Remove.
14973 (align_functions_max_skip): Remove.
14974 (align_loops_value): Remove.
14975 (align_jumps_value): Remove.
14976 (align_labels_value): Remove.
14977 (align_functions_value): Remove.
14978 * output.h (label_to_alignment): Change return type to align_flags.
14979 (label_to_max_skip): Remove.
14980 * target.def: Remove loop_align_max_skip, label_align_max_skip,
14981 jump_align_max_skip macros.
14982 * targhooks.h (default_loop_align_max_skip): Remove.
14983 (default_label_align_max_skip): Likewise.
14984 (default_jump_align_max_skip): Likewise.
14985 (default_label_align_after_barrier_max_skip): Remove.
14986 * toplev.c (read_log_maxskip): Use ::normalize function.
14987 (parse_N_M): Remove not used argument and also call ::normalize.
14988 (parse_alignment_opts): Do not pass unused arguments.
14989 * varasm.c (assemble_start_function): Use directly align_functions
14990 instead of removed macros.
14991 * system.h: Do not poison removed macros.
14993 2018-07-17 Jakub Jelinek <jakub@redhat.com>
14995 PR middle-end/86539
14996 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
14997 and cond temporaries don't have reference type if iterator has
14998 pointer type. For init use &for_pre_body instead of pre_p if
14999 for_pre_body is non-empty.
15001 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
15003 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
15004 double-double modes to SFmode directly directly.
15005 (trunc<mode>sf2_fprs): Delete.
15007 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
15009 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
15010 for conversions between IFmode and the decimal floating point modes.
15011 (init_float128_ieee): Use the correct names for conversions between
15012 KFmode and the decimal floating point modes.
15014 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
15016 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
15017 for the conversions between TDmode and IFmode.
15018 (init_float128_ieee): Use more correct names for the conversions
15019 between TDmode and KFmode.
15021 2018-07-16 Jakub Jelinek <jakub@redhat.com>
15023 PR tree-optimization/86526
15024 * builtins.c (expand_builtin_memcmp): Formatting fixes.
15025 (inline_expand_builtin_string_cmp): Likewise.
15026 (inline_string_cmp): Likewise. Use c_readstr instead of
15027 builtin_memcpy_read_str. Add unit_mode temporary.
15029 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
15031 PR middle-end/86528
15032 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
15033 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
15035 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
15037 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
15038 Alphabetize prototypes of built-in functions, separating out
15039 built-in functions that are listed in this section but should be
15040 described elsewhere.
15042 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
15045 * expmed.c (emit_store_flag): Do not emit setcc followed by a
15046 conditional move when trapping comparison was split to a
15047 non-trapping one (and vice versa).
15049 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
15051 * config/s390/s390.c (s390_function_profiler): Generate nops
15052 instead of profiler call sequences.
15053 * config/s390/s390.opt: Add the new option.
15055 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
15057 * config/s390/s390.c (s390_function_profiler): Generate
15058 __mcount_loc section.
15059 * config/s390/s390.opt: Add the new option.
15061 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
15063 * common.opt: Add the new warning.
15064 * config/s390/s390.c (s390_function_profiler): Emit "brasl
15065 %r0,__fentry__" when -mfentry is specified.
15066 (s390_option_override_internal): Disallow -mfentry for 31-bit
15068 * config/s390/s390.opt: Add the new option.
15070 2018-07-16 Richard Biener <rguenther@suse.de>
15073 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
15074 for function-local FUNCTION_DECL and RESULT_DECL immediately.
15076 2018-07-16 Martin Liska <mliska@suse.cz>
15079 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
15082 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
15084 * config/arc/arcHS.md: Update ARCHS scheduling rules.
15086 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
15088 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
15090 * config/arc/arc-cpus.def (hs4x): New cpu.
15092 * config/arc/arc-tables.opt: Regenerate.
15093 * config/arc/arc.c (arc_sched_issue_rate): New function.
15094 (TARGET_SCHED_ISSUE_RATE): Define.
15095 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
15096 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
15098 (attr tune): Add ARCHS4x tune values.
15099 (attr tune_dspmpy): Define.
15100 (*tst): Correct instruction type.
15101 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
15102 * config/arc/arcHS4x.md: New file.
15103 * config/arc/fpu.md: Update instruction type attributes.
15104 * config/arc/t-multilib: Regenerate.
15106 2018-07-16 Tom de Vries <tdevries@suse.de>
15109 * var-tracking.c (vt_initialize): Fix pre_dec handling.
15111 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15113 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
15116 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
15118 PR tree-optimization/86514
15119 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
15120 conversion to a boolean type from a type with greater precision.
15122 2018-07-16 Tom de Vries <tdevries@suse.de>
15124 * var-tracking.c (vt_initialize): Print adjusted insn slim if
15125 dump_flags request TDF_SLIM.
15127 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
15129 * fold-const.c (int_const_binop_1): Abstract...
15130 (wide_int_binop): ...wide int code here.
15131 (poly_int_binop): ...poly int code here.
15132 Abstract the rest of int_const_binop_1 into int_const_binop.
15133 * fold-const.h (wide_int_binop): New.
15134 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
15135 Remove useless PLUS/MINUS_EXPR case.
15136 (zero_nonzero_bits_from_vr): Move wide int code...
15137 (zero_nonzero_bits_from_bounds): ...here.
15138 (extract_range_from_binary_expr_1): Move mask optimization code...
15139 (range_easy_mask_min_max): ...here.
15140 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
15141 (range_easy_mask_min_max): New.
15143 2018-07-15 Jeff Law <law@redhat.com>
15146 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
15149 2018-07-14 Jim Wilson <jimw@sifive.com>
15151 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
15153 2018-07-14 Paul Koning <ni1d@arrl.net>
15155 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
15157 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
15159 * lto-streamer-out.c (copy_function_or_variable): Dump info about
15162 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
15163 Steve Munroe <munroesj52@gmail.com>
15165 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
15166 (_mm_andnot_si128): Likewise.
15167 (_mm_or_si128): Likewise.
15168 (_mm_xor_si128): Likewise.
15170 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
15172 PR middle-end/78809
15173 * builtins.c (expand_builtin_memcmp): Inline the calls first
15174 when result_eq is false.
15175 (expand_builtin_strcmp): Inline the calls first.
15176 (expand_builtin_strncmp): Likewise.
15177 (inline_string_cmp): New routine. Expand a string compare
15178 call by using a sequence of char comparison.
15179 (inline_expand_builtin_string_cmp): New routine. Inline expansion
15180 a call to str(n)cmp/memcmp.
15181 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
15183 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
15185 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
15187 * config/arm/driver-arm.c: Include arm-native.h.
15188 (host_detect_local_cpu): Use auto-generated data tables.
15189 (vendors, arm_cpu_table): Delete. Move part information to ...
15190 * config/arm/arm-cpus.in: ... here.
15191 * config/arm/parsecpu.awk (gen_native): New function.
15192 (vendor, part): New CPU fields.
15193 (END): Add support for building the native CPU detection tables.
15194 * config/arm/t-arm (arm-native.h): Add build rule.
15195 (driver-arm.o): Add dependency on arm-native.h.
15197 2018-07-13 Richard Biener <rguenther@suse.de>
15199 PR middle-end/85974
15200 * match.pd (addr1 - addr2): Allow either of the operand to
15203 2018-07-13 Tom de Vries <tdevries@suse.de>
15205 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
15208 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
15210 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
15211 arrays instead of numbered variables.
15213 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
15215 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
15216 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
15218 2018-07-13 Richard Biener <rguenther@suse.de>
15221 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
15222 instead of get_context_die.
15224 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15225 Richard Biener <rguenther@suse.de>
15227 PR middle-end/86489
15228 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
15229 that the loop latch destination where phi is defined.
15231 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
15233 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
15234 (riscv_expand_epilogue): Add assertion to check interrupt mode.
15235 (riscv_set_current_function): Extract getting interrupt type to new
15237 (riscv_get_interrupt_type): New function.
15238 (riscv_merge_decl_attributes): New function, checking interrupt type is
15240 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
15242 2018-07-12 Paul Koning <ni1d@arrl.net>
15244 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
15247 2018-07-12 Paul Koning <ni1d@arrl.net>
15249 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
15250 zero reference, add doloop_end instead.
15251 * doc/md.texi (decrement_and_branch_until_zero): Remove.
15252 (Looping patterns): Remove decrement_and_branch_until_zero. Add
15253 detail for doloop_end.
15255 2018-07-12 Martin Sebor <msebor@redhat.com>
15258 * attribs.c (decl_attributes): Reject conflicting attributes before
15259 calling attribute handlers.
15261 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
15263 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
15265 (gcc::dump_manager::get_dump_file_name): likewise.
15266 (dump_begin): Likewise.
15267 * dumpfile.h (dump_begin): Update prototype.
15268 (gcc::dump_manager::get_dump_file_name,
15269 gcc::dump_manager::get_dump_file_name): Update prototype.
15271 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
15273 * internal-fn.h (vectorizable_internal_fn_p): New function.
15274 * tree-vect-slp.c (compatible_calls_p): Likewise.
15275 (vect_build_slp_tree_1): Remove nops argument. Handle calls
15276 to internal functions.
15277 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
15279 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
15281 * fold-const.h (inverse_conditions_p): Declare.
15282 * fold-const.c (inverse_conditions_p): New function.
15283 * match.pd: Use inverse_conditions_p. Add folds of view_converts
15284 that test the inverse condition of a conditional internal function.
15285 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
15286 * internal-fn.c (internal_fn_mask_index): Handle conditional
15287 internal functions.
15288 (vectorized_internal_fn_supported_p): New function.
15289 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
15290 (any_pred_load_store): Replace with...
15291 (need_to_predicate): ...this new variable.
15292 (redundant_ssa_names): New variable.
15293 (ifcvt_can_use_mask_load_store): Move initial checks to...
15294 (ifcvt_can_predicate): ...this new function. Handle tree codes
15295 for which a conditional internal function exists.
15296 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
15297 instead of ifcvt_can_use_mask_load_store. Update after variable
15299 (predicate_load_or_store): New function, split out from
15300 predicate_mem_writes.
15301 (check_redundant_cond_expr): New function.
15302 (value_available_p): Likewise.
15303 (predicate_rhs_code): Likewise.
15304 (predicate_mem_writes): Rename to...
15305 (predicate_statements): ...this. Use predicate_load_or_store
15306 and predicate_rhs_code.
15307 (combine_blocks, tree_if_conversion): Update after above name changes.
15308 (ifcvt_local_dce): Handle redundant_ssa_names.
15309 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
15310 general conditional functions.
15311 * tree-vect-stmts.c (vectorizable_call): Likewise.
15313 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
15314 Alan Hayward <alan.hayward@arm.com>
15315 David Sherwood <david.sherwood@arm.com>
15317 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
15318 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
15319 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
15320 plus and minus and convert them into IFN_COND_FMA-based sequences.
15321 (convert_mult_to_fma): Handle conditional plus and minus.
15323 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
15325 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
15326 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
15327 (cond_fnms_optab): New optabs.
15328 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
15329 internal functions.
15330 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
15331 * internal-fn.h (get_conditional_internal_fn): Declare.
15332 (get_unconditional_internal_fn): Likewise.
15333 * internal-fn.c (cond_ternary_direct): New macro.
15334 (expand_cond_ternary_optab_fn): Likewise.
15335 (direct_cond_ternary_optab_supported_p): Likewise.
15336 (FOR_EACH_COND_FN_PAIR): Likewise.
15337 (get_conditional_internal_fn): New function.
15338 (get_unconditional_internal_fn): Likewise.
15339 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
15340 (gimple_match_op::gimple_match_op): Add a new overload for 5
15342 (gimple_match_op::set_op): Likewise.
15343 (gimple_resimplify5): Declare.
15344 * genmatch.c (decision_tree::gen): Generate simplifications for
15346 * gimple-match-head.c (gimple_simplify): Define an overload for
15347 5 operands. Handle calls with 5 arguments in the top-level overload.
15348 (convert_conditional_op): Handle conversions from unconditional
15349 internal functions to conditional ones.
15350 (gimple_resimplify5): New function.
15351 (build_call_internal): Pass a fifth operand.
15352 (maybe_push_res_to_seq): Likewise.
15353 (try_conditional_simplification): Try converting conditional
15354 internal functions to unconditional internal functions.
15355 Handle 3-operand unconditional forms.
15356 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
15357 Define ternary equivalents of the current rules for binary conditional
15358 internal functions.
15359 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
15360 ternary operations.
15361 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
15362 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
15363 (optab): Handle them.
15364 (SVE_COND_FP_TERNARY): New int iterator.
15365 (sve_fmla_op, sve_fmad_op): New int attributes.
15366 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
15367 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
15368 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
15370 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
15372 * target.def (preferred_else_value): New target hook.
15373 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
15374 * doc/tm.texi: Regenerate.
15375 * targhooks.h (default_preferred_else_value): Declare.
15376 * targhooks.c (default_preferred_else_value): New function.
15377 * internal-fn.h (conditional_internal_fn_code): Declare.
15378 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
15379 (get_conditional_internal_fn): Use it.
15380 (conditional_internal_fn_code): New function.
15381 * gimple-match.h (gimple_match_cond): New struct.
15382 (gimple_match_op): Add a cond member function.
15383 (gimple_match_op::gimple_match_op): Update all forms to take a
15385 * genmatch.c (expr::gen_transform): Use the same condition as res_op
15386 for the suboperation, but don't specify a particular else_value.
15387 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
15388 (visit_nary_op, visit_reference_op_load): Pass
15389 gimple_match_cond::UNCOND to the gimple_match_op constructor.
15390 * gimple-match-head.c: Include tree-eh.h
15391 (convert_conditional_op): New function.
15392 (maybe_resimplify_conditional_op): Likewise.
15393 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
15394 (gimple_resimplify2): Likewise.
15395 (gimple_resimplify3): Likewise.
15396 (gimple_resimplify4): Likewise.
15397 (maybe_push_res_to_seq): Return null for conditional operations.
15398 (try_conditional_simplification): New function.
15399 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
15401 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
15403 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
15405 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
15407 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
15409 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
15411 (hash_tree): Do not hash DECL_FCONTEXT
15412 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
15413 Do not stream DECL_FCONTEXT.
15414 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
15415 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
15417 2018-07-12 Richard Biener <rguenther@suse.de>
15420 * dwarf2out.c (gen_block_die): Only output blocks when they have
15421 at least one !DECL_IGNORED_P variable.
15423 2018-07-12 Richard Biener <rguenther@suse.de>
15426 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
15427 Remove -mieee-fp handling.
15429 2018-07-12 Richard Biener <rguenther@suse.de>
15431 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
15432 left-over from last patch.
15434 2018-07-12 Jakub Jelinek <jakub@redhat.com>
15436 PR tree-optimization/86492
15437 * gimple-ssa-store-merging.c
15438 (imm_store_chain_info::coalesce_immediate_stores): Call
15439 check_no_overlap even for the merge_overlapping case. Formatting fix.
15441 2018-07-12 Richard Biener <rguenther@suse.de>
15443 PR middle-end/86479
15444 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
15445 move possibly trapping operations into the conditional.
15447 2018-07-12 Richard Biener <rguenther@suse.de>
15449 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
15450 (vn_lookup_simplify_result): Remove recursion limit applied
15452 (vn_nary_build_or_lookup_1): Adjust.
15453 (try_to_simplify): Likewise.
15454 * gimple-match-head.c (gimple_resimplify1): Instead apply one
15456 (gimple_resimplify2): Likewise.
15457 (gimple_resimplify3): Likewise.
15458 (gimple_resimplify4): Likewise.
15460 2018-07-11 Jakub Jelinek <jakub@redhat.com>
15462 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
15463 Use __mmask64 type instead of __mmask8 for __M argument.
15464 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
15465 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
15467 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
15468 __mmask16 for __M argument.
15469 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
15470 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
15471 to __mmask16 instead of __mmask8.
15472 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
15473 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
15474 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
15475 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
15476 instead of __mmask16 for __U argument.
15477 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
15478 __mmask16 instead of __mmask8 for __U argument.
15479 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
15481 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
15483 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
15485 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
15487 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
15489 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
15491 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
15493 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
15495 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
15497 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
15499 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
15500 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
15501 return type as well as __M argument type and all casts from __mmask8
15503 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
15504 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
15505 return type as well as __M argument type and all casts from __mmask8
15507 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
15508 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
15509 return type as well as __M argument type and all casts from __mmask8
15511 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
15512 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
15513 return type as well as __M argument type and all casts from __mmask8
15515 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
15516 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
15519 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
15521 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
15522 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
15525 2018-07-11 Paul Koning <ni1d@arrl.net>
15527 * doc/md.texi (define_subst): Document how multiple occurrences of
15528 the same argument in the replacement pattern are handled.
15530 2018-07-11 Paul Koning <ni1d@arrl.net>
15532 * doc/extend.texi (Common Variable Attributes): Move "mode" into
15533 alphabetical order.
15534 (Common Type Attributes): Add "mode" attribute.
15536 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
15538 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
15539 stream DECL_ORIGINAL_TYPE.
15540 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
15541 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
15542 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
15543 Do not walk original type.
15544 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
15546 (write_ts_decl_non_common_tree_pointers): Do not stream
15548 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
15549 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
15551 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
15553 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
15554 thread twice from the same starting edge.
15556 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
15558 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
15559 * gimple.c (gimple_call_nonnull_result_p): ...here...
15560 (gimple_call_nonnull_arg): ...and here.
15561 * gimple.h (gimple_call_nonnull_result_p): New.
15562 (gimple_call_nonnull_arg): New.
15564 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
15566 * config/arm/arm-cpus.in: Move information from fpu field of each
15567 cpu definition to the isa field.
15568 * config/arm/parsecpu.awk (fpu): Delete match rule.
15569 (gen_comm_data): Don't add bits from the CPU's FPU entry.
15571 2018-07-11 Richard Biener <rguenther@suse.de>
15574 * dwarf2out.c (init_sections_and_labels): Use
15575 output_asm_line_debug_info consistently.
15576 (dwarf2out_early_finish): Likewise.
15577 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
15580 2018-07-11 Richard Biener <rguenther@suse.de>
15582 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
15583 Rework father_bb setting in a way to avoid propagating constants
15584 multiple times on a loop body.
15586 2018-07-10 Mark Wielaard <mark@klomp.org>
15589 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
15591 2018-07-10 Richard Biener <rguenther@suse.de>
15593 * hash-map.h (hash_map::iterator::operator*): Return
15594 references to key and value.
15596 2018-07-10 Jakub Jelinek <jakub@redhat.com>
15599 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
15600 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
15601 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
15602 (gimplify_omp_for): For composite loops, move outer
15603 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
15604 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
15605 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
15606 TREE_LIST for both the original class iterator and the "last" helper
15607 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
15608 loop, remember has_decl_expr from outer composite loops for the
15609 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
15611 2018-07-09 Martin Sebor <msebor@redhat.com>
15613 PR middle-end/77357
15614 PR middle-end/86428
15615 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
15616 accessing implicitly initialized array elements.
15617 * expr.c (string_constant): Handle string initializers of
15618 character arrays within aggregates.
15619 * gimple-fold.c (fold_array_ctor_reference): Add argument.
15620 Store element offset. As a special case, handle zero size.
15621 (fold_nonarray_ctor_reference): Same.
15622 (fold_ctor_reference): Add argument. Store subobject offset.
15623 * gimple-fold.h (fold_ctor_reference): Add argument.
15625 2018-07-09 Paul Koning <ni1d@arrl.net>
15627 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
15628 (pdp11_insn_cost): New function.
15629 (pdp11_md_asm_adjust): New function.
15630 (TARGET_INVALID_WITHIN_DOLOOP): Define.
15631 (pdp11_rtx_costs): Update to match machine better.
15632 (output_addr_const_pdp11): Correct format mismatch warnings.
15633 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
15634 * config/pdp11/pdp11.md: General change to add base_cost and/or
15635 length attributes for use by new pdp11_insn_cost function.
15636 (MIN_BRANCH): Correct definition.
15638 (doloop_end): Use standard pattern name for looping pattern.
15639 (doloop_end_nocc): New.
15640 (movsf): Add another constraint alternative.
15641 (zero_extendqihi2): Add constraint alternatives for not in place
15643 (zero_extendhisi2): Remove.
15644 (shift patterns): Add CC handling variants.
15648 (define_peephole2): New peephole to recognize mov that sets CC for
15651 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15653 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
15654 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
15657 2018-07-09 Jakub Jelinek <jakub@redhat.com>
15660 * real.c (real_nextafter): Return true if result is denormal.
15662 2018-07-09 Martin Liska <mliska@suse.cz>
15664 * common.opt: Add back wrongly removed attribute.
15666 2018-07-09 Richard Biener <rguenther@suse.de>
15669 * dwarf2out.c (gen_block_die): For an early generated DIE
15670 always output high/low PC attributes.
15672 2018-07-09 Tom de Vries <tdevries@suse.de>
15674 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
15675 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
15676 onto VAR_DECL with abstract origin.
15678 2018-07-07 Jim Wilson <jimw@sifive.com>
15680 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
15682 2018-07-07 Tom de Vries <tdevries@suse.de>
15684 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
15686 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
15688 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
15690 (combine_bound): Use wide-int overflow calculation instead of
15692 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
15694 * fold-const.c (int_const_binop_2): Same.
15695 (extract_muldiv_1): Same.
15696 (fold_div_compare): Same.
15697 (fold_abs_const): Same.
15699 * poly-int.h (add): Same.
15703 * predict.c (predict_iv_comparison): Same.
15704 * profile-count.c (slow_safe_scale_64bit): Same.
15705 * simplify-rtx.c (simplify_const_binary_operation): Same.
15706 * tree-chrec.c (tree_fold_binomial): Same.
15707 * tree-data-ref.c (split_constant_offset_1): Same.
15708 * tree-if-conv.c (idx_within_array_bound): Same.
15709 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
15710 * tree-ssa-phiopt.c (minmax_replacement): Same.
15711 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
15712 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
15713 * vr-values.c (vr_values::adjust_range_with_scev): Same.
15714 * wide-int.cc (wi::add_large): Same.
15715 (wi::mul_internal): Same.
15716 (wi::sub_large): Same.
15717 (wi::divmod_internal): Same.
15718 * wide-int.h: Change overflow type to overflow_type for neg, add,
15719 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
15720 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
15721 mul_internal, divmod_internal.
15722 (overflow_type): New enum.
15723 (accumulate_overflow): New.
15725 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
15727 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
15728 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
15730 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
15732 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
15733 argument is checked for zero before entering loop, avoid checking again.
15735 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
15737 * gimplify.h (generic_expr_could_trap_p): Set as global function.
15738 * gimplify.h (generic_expr_could_trap_p): Likwise.
15739 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
15741 2018-07-06 Jakub Jelinek <jakub@redhat.com>
15743 PR tree-optimization/86401
15744 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
15745 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
15746 (fold_bit_and_mask): ... here. New helper function for match.pd.
15747 * fold-const.h (fold_bit_and_mask): Declare.
15748 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
15750 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
15753 * target.def (translate_mode_attribute): New hook.
15754 * targhooks.h (default_translate_mode_attribute): Declare.
15755 * targhooks.c (default_translate_mode_attribute): New function.
15756 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
15757 * doc/tm.texi: Regenerate.
15758 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
15759 (rs6000_translate_mode_attribute): New function.
15761 2018-07-06 Paul Koning <ni1d@arrl.net>
15763 * doc/md.texi (define_split): Document DONE and FAIL.
15764 (define_peephole2): Ditto.
15766 2018-07-05 Jeff Law <law@redhat.com>
15768 PR tree-optimization/86010
15769 * tree-ssa-dse.c (compute_trims): More aggressively trim at
15770 both the head and tail of mem* and str* calls.
15772 2018-07-05 Jim Wilson <jimw@sifive.com>
15774 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
15776 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
15778 * config/aarch64/aarch64-simd.md: correct flags text for
15779 MIN_EXPR replacement.
15781 2018-07-05 James Clarke <jrtc27@jrtc27.com>
15783 * configure: Regenerated.
15785 2018-07-05 Carl Love <cel@us.ibm.com>
15787 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
15788 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
15789 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
15790 VSX_BUILTIN_DOUBLEL_V4SF.
15792 2018-07-05 Martin Sebor <msebor@redhat.com>
15795 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
15796 than its domain to compute its the upper bound of a char array.
15798 2018-07-05 Nathan Sidwell <nathan@acm.org>
15800 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
15801 * doc/cpp.texi: Update comment.
15802 * doc/tm.texi: Rebuilt.
15803 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
15804 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
15805 * doc/extend.texi (Backwards Compatibility): Clarify it is system
15806 headers affected by extern "C".
15807 * system.h: Poison NO_IMPLICIT_EXTERN_C.
15808 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
15809 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
15810 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
15811 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
15812 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
15813 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
15814 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
15815 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
15816 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
15817 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
15818 config/sparc/sp64-elf.h, config/spu/spu.h,
15819 config/stormy16/stormy16.h, config/v850/v850.h,
15820 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
15821 define NO_IMPLICIT_EXTERN_C.
15822 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
15824 2018-07-05 Tamar Christina <tamar.christina@arm.com>
15827 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
15828 instead of GET_MODE_SIZE when comparing Units.
15830 2018-07-05 Tamar Christina <tamar.christina@arm.com>
15833 * rtlanal.c (set_noop_p): Constrain on mode change,
15834 include hard-reg-set.h
15836 2018-07-05 Tamar Christina <tamar.christina@arm.com>
15838 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
15840 2018-07-05 Jakub Jelinek <jakub@redhat.com>
15843 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
15846 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
15848 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
15850 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
15853 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
15855 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
15857 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
15859 PR middle-end/86380
15860 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
15862 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
15864 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
15865 neg_*_op* variables.
15867 2018-07-04 Martin Liska <mliska@suse.cz>
15869 * tree-switch-conversion.c: Define
15870 max_ratio_for_speed and max_ratio_for_size constants.
15872 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
15873 Martin Liska <mliska@suse.cz>
15875 PR middle-end/66240
15878 * common.opt: Rename align options with 'str_' prefix.
15879 * common/config/i386/i386-common.c (set_malign_value): New
15881 (ix86_handle_option): Use it to set -falign-* options/
15882 * config/aarch64/aarch64-protos.h (struct tune_params): Change
15883 type from int to string.
15884 * config/aarch64/aarch64.c: Update default values from int
15886 * config/alpha/alpha.c (alpha_override_options_after_change):
15888 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
15889 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15890 max skip conditionally.
15891 * config/i386/freebsd.h (SUBALIGN_LOG): New.
15892 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15893 max skip conditionally.
15894 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15895 max skip conditionally.
15896 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
15897 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15898 max skip conditionally.
15899 * config/i386/i386.c (struct ptt): Change type from int to
15901 (ix86_default_align): Set default values.
15902 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
15903 max skip conditionally.
15904 * config/i386/iamcu.h (SUBALIGN_LOG): New.
15905 (ASM_OUTPUT_MAX_SKIP_ALIGN):
15906 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
15907 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15908 max skip conditionally.
15909 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
15910 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
15911 * config/i386/x86-64.h (SUBALIGN_LOG): New.
15912 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15913 max skip conditionally.
15914 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
15915 * config/ia64/ia64.c (ia64_option_override): Set default values
15916 for alignment options.
15917 * config/m68k/m68k.c: Handle new str_align_* options.
15918 * config/mips/mips.c (mips_set_compression_mode): Change
15920 (mips_option_override): Set default values for options.
15921 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
15923 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15925 * config/rx/rx.c (rx_option_override): Likewise.
15926 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
15927 (LABEL_ALIGN): Use align_labels_log.
15928 (LOOP_ALIGN): Use align_loops_align.
15929 * config/s390/s390.c (s390_asm_output_function_label): Use new
15931 * config/sh/sh.c (sh_override_options_after_change):
15932 Change type of constants.
15933 * config/spu/spu.c (spu_sched_init): Likewise.
15934 * config/sparc/sparc.c (sparc_option_override): Set default
15935 values for options.
15936 * config/visium/visium.c (visium_option_override): Likewise.
15937 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
15938 emit p2align format with last argument if it's not needed.
15939 * doc/invoke.texi: Document extended format of -falign-*.
15940 * final.c: Use align_labels alignment.
15941 * flags.h (struct target_flag_state): Change type to use
15943 (struct align_flags_tuple): New.
15944 (struct align_flags): Likewise.
15945 (align_loops_log): Redefine macro to use new types.
15946 (align_loops_max_skip): Redefine macro to use new types.
15947 (align_jumps_log): Redefine macro to use new types.
15948 (align_jumps_max_skip): Redefine macro to use new types.
15949 (align_labels_log): Redefine macro to use new types.
15950 (align_labels_max_skip): Redefine macro to use new types.
15951 (align_functions_log): Redefine macro to use new types.
15952 (align_loops): Redefine macro to use new types.
15953 (align_jumps): Redefine macro to use new types.
15954 (align_labels): Redefine macro to use new types.
15955 (align_functions): Redefine macro to use new types.
15956 (align_functions_max_skip): Redefine macro to use new types.
15957 (align_loops_value): New macro.
15958 (align_jumps_value): New macro.
15959 (align_labels_value): New macro.
15960 (align_functions_value): New macro.
15961 * function.c (invoke_set_current_function_hook): Propagate
15962 alignment values from flags to global variables default in
15964 * ipa-icf.c (sem_function::equals_wpa): Use
15965 cl_optimization_option_eq instead of memcmp.
15966 * lto-streamer.h (cl_optimization_stream_out): Support streaming
15968 (cl_optimization_stream_in): Likewise.
15969 * optc-save-gen.awk: Support strings in cl_optimization.
15970 * opth-gen.awk: Likewise.
15971 * opts.c (finish_options): Remove error checking of invalid
15973 (MAX_CODE_ALIGN): Remove.
15974 (MAX_CODE_ALIGN_VALUE): Likewise.
15975 (parse_and_check_align_values): New function.
15976 (check_alignment_argument): Likewise.
15977 (common_handle_option): Use check_alignment_argument.
15978 * opts.h (parse_and_check_align_values): Declare.
15979 * toplev.c (init_alignments): Remove.
15980 (read_log_maxskip): New.
15981 (parse_N_M): Likewise.
15982 (parse_alignment_opts): Likewise.
15983 (backend_init_target): Remove usage of init_alignments.
15984 * toplev.h (parse_alignment_opts): Declare.
15985 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
15987 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
15988 * tree.c (cl_option_hasher::equal): New.
15989 * varasm.c: Use new global macros.
15991 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
15993 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
15994 Use a simpler align directive also if MAXSKIP = ALIGN-1.
15995 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15996 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15997 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15998 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
15999 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
16000 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
16001 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
16002 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
16003 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
16004 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
16005 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
16007 2018-07-04 Martin Liska <mliska@suse.cz>
16008 Jonathan Wakely <jwakely@redhat.com>
16010 * coverage.c: Use correct type.
16011 * doc/invoke.texi: Language correction.
16013 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
16016 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
16017 ENDBRANCH for non-tail call which may return via indirect branch.
16018 * doc/extend.texi: Document indirect_return attribute.
16020 2018-07-03 Martin Sebor <msebor@redhat.com>
16022 PR tree-optimization/86274
16023 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
16025 (format_floating): Correct handling of infinities and NaNs.
16027 2018-07-03 Martin Sebor <msebor@redhat.com>
16029 * print-tree.c (print_real_cst): New function.
16030 (print_node_brief): Call it.
16031 (print_node): Ditto.
16033 2018-07-03 Jeff Law <law@redhat.com>
16035 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
16036 into a single pattern.
16038 * config/h8300/h8300.md (ors code_iterator): New.
16039 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
16040 a single pattern and single splitter.
16041 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
16042 (iorqi3_1, xorqi3_1): Likewise.
16043 (iorqi3, xorqi3 expanders): Similarly.
16045 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
16046 (movmd_internal) into a single pattern using the P mode iterator.
16047 (movmd splitters): Similarly.
16048 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
16049 (movsd splitters): Similarly.
16051 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
16052 ADDB, ADDW and ADDL into a single ADD attribute which selects the
16053 right table based on the size of the operand.
16054 * config/h8300/h8300.md (length_table): Corresponding changes. All
16055 references to "addb", "addw" and "addl" changed to "add".
16056 (btst patterns): Merge two variants into a single pattern.
16057 (tstqi, tsthi): Likewise.
16058 (addhi3_incdec, addsi3_incdec): Likewise.
16059 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
16060 (mulhi3, mulsi3): Likewise.
16061 (udivhi3, udivsi3): Likewise.
16062 (divhi3, divsi3): Likewise.
16063 (andorqi3, andorhi3, andorsi3): Likewise.
16065 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
16068 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
16069 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
16071 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16073 PR tree-optimization/85694
16074 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
16075 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
16077 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
16078 (<u>avg<mode>3_ceil): New patterns.
16080 2018-07-03 David Malcolm <dmalcolm@redhat.com>
16082 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
16083 scan-tree-dump directive.
16084 * gcc.dg/vect/slp-perm-2.c: Likewise.
16085 * gcc.dg/vect/slp-perm-3.c: Likewise.
16086 * gcc.dg/vect/slp-perm-5.c: Likewise.
16087 * gcc.dg/vect/slp-perm-6.c: Likewise.
16088 * gcc.dg/vect/slp-perm-7.c: Likewise.
16089 * gcc.dg/vect/slp-perm-8.c: Likewise.
16091 2018-07-03 Marek Polacek <polacek@redhat.com>
16093 PR middle-end/86202
16094 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
16096 2018-07-03 Richard Biener <rguenther@suse.de>
16099 * tree-ssa-structalias.c (find_func_clobbers): Properly
16100 handle indirect calls.
16102 2018-07-03 Jeff Law <law@redhat.com>
16104 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
16105 (shifts): New code iterator.
16106 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
16107 expander. Fix HImode handling on H8/SX.
16108 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
16109 (subqi3, subhi3, subsi3 expanders): Likewise.
16110 (andqi3, andhi3, andsi3 expanders): Likewise.
16111 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
16112 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
16113 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
16114 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
16115 (zero_extendqihi2, zero_extendqisi2): Likewise.
16116 (extendqihi2, extendqisi2): Likewise.
16117 (rotlqi3, rotlhi3, rotlsi3): Likewise.
16118 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
16119 (rotlqi3_1, rotlhi3_1): Likewise.
16120 (logicalhi3_sn, logicalsi3_sn): Likewise.
16121 (logicalhi3, logicalsi3): Likewise.
16123 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16125 * tree-vect-patterns.c (vect_recog_rotate_pattern)
16126 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
16127 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
16128 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
16129 type to append_pattern_def_seq instead of creating a stmt_vec_info
16131 (build_mask_conversion): Likewise. Remove vinfo argument.
16132 (vect_add_conversion_to_patterm): Likewise, renaming to...
16133 (vect_add_conversion_to_pattern): ...this.
16134 (vect_recog_mask_conversion_pattern): Update call to
16135 build_mask_conversion. Pass the vector type to
16136 append_pattern_def_seq here too.
16137 (vect_recog_gather_scatter_pattern): Update call to
16138 vect_add_conversion_to_pattern.
16140 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16142 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
16143 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
16144 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
16145 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
16146 STMT_VINFO_PATTERN_DEF_SEQ to null here.
16147 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
16148 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
16149 append_pattern_def_seq instead of new_pattern_def_seq.
16150 (vect_recog_divmod_pattern): Do both of the above.
16151 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
16154 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16156 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
16157 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
16158 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
16159 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
16160 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
16161 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
16162 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
16163 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
16164 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
16165 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
16166 parameter with a single stmt_vec_info.
16167 (vect_recog_func_ptr): Likewise.
16168 (vect_recog_gather_scatter_pattern): Likewise, folding in...
16169 (vect_try_gather_scatter_pattern): ...this.
16170 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
16171 the stmt_vec_info of the statement to be matched. Don't clear
16172 STMT_VINFO_RELATED_STMT.
16173 (vect_pattern_recog): Update call accordingly.
16175 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16177 PR tree-optimization/85694
16178 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
16179 (uavgM3_ceil): Document new optabs.
16180 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
16181 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
16183 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
16184 (savg_ceil_optab): New optabs.
16185 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
16186 (vect_vect_recog_func_ptrs): Add it.
16187 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
16188 constant directly from the associated lhs.
16190 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16192 * tree-vect-patterns.c (vect_split_statement): New function.
16193 (vect_convert_input): Use it to try to split an existing cast.
16195 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16197 * poly-int.h (print_hex): New function.
16198 * dumpfile.h (dump_dec, dump_hex): Declare.
16199 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
16200 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
16201 min_input_precision, operation_precision and operation_sign.
16202 * tree-vect-patterns.c (vect_get_range_info): New function.
16203 (vect_same_loop_or_bb_p, vect_single_imm_use)
16204 (vect_operation_fits_smaller_type): Delete.
16205 (vect_look_through_possible_promotion): Add an optional
16206 single_use_p parameter.
16207 (vect_recog_over_widening_pattern): Rewrite to use new
16208 stmt_vec_info infomration. Handle one operation at a time.
16209 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
16210 (vect_truncatable_operation_p, vect_set_operation_type)
16211 (vect_set_min_input_precision): New functions.
16212 (vect_determine_min_output_precision_1): Likewise.
16213 (vect_determine_min_output_precision): Likewise.
16214 (vect_determine_precisions_from_range): Likewise.
16215 (vect_determine_precisions_from_users): Likewise.
16216 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
16217 (vect_vect_recog_func_ptrs): Put over_widening first.
16219 (vect_pattern_recog): Call vect_determine_precisions.
16221 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16223 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
16224 statements that have been replaced by further pattern statements.
16225 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
16227 2018-07-03 Richard Biener <rguenther@suse.de>
16229 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
16230 always set *dt. Dump vectype in vectype overload.
16231 * dumpfile.h (dump_gimple_expr): New function.
16232 (dump_gimple_expr_loc): Likewise.
16233 * dumpfile.c (dump_gimple_expr): New function.
16234 (dump_gimple_expr_loc): Likewise.
16236 2018-07-02 Jeff Law <law@redhat.com>
16238 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
16239 the H8/300, H8/300H and H8/S variants into a single pattern.
16240 (movhi_h8300, movqi_h8300hs): Similarly.
16241 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
16242 (QHI mode iterator): New.
16244 * config/h8300/h8300.md: Remove trailing whitespace.
16246 2018-07-02 Jim Wilson <jimw@sifive.com>
16248 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
16249 instead of emit_insn for interrupt returns.
16250 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
16251 (riscv_sret, riscv_uret): Likewise.
16253 2018-07-02 David Malcolm <dmalcolm@redhat.com>
16255 * pretty-print.c (selftest::test_pp_format): Move save and restore
16256 of quotes to class auto_fix_quotes, and add an instance.
16257 * selftest.c: Include "intl.h".
16258 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
16259 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
16260 * selftest.h (selftest::auto_fix_quotes): New class.
16262 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
16264 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
16265 (aarch64_sve_prepare_conditional_op): Remove.
16266 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
16267 Allow aarch64_simd_reg_or_zero as select operand; remove
16268 the aarch64_sve_prepare_conditional_op call.
16269 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
16270 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
16271 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
16272 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
16273 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
16274 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
16275 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
16276 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
16277 and a splitters to match all of the *_any patterns.
16278 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
16280 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
16281 (SVE_COND_FP_BINARY_REV): Remove.
16282 (sve_int_op_rev, sve_fp_op_rev): New.
16283 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
16284 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
16285 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
16286 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
16287 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
16288 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
16289 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
16290 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
16291 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
16293 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
16294 Remove match_dup 1 from the inner unspec.
16295 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
16297 * config/aarch64/aarch64.md (movprfx): New attr.
16298 (length): Default movprfx to 8.
16299 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
16300 (*madd<SVE_I>, *msub<SVE_I): Likewise.
16301 (*<su>mul<SVE_I>3_highpart): Likewise.
16302 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
16303 (*v<ASHIFT><SVE_I>3): Likewise.
16304 (*<su><MAXMIN><SVE_I>3): Likewise.
16305 (*<su><MAXMIN><SVE_F>3): Likewise.
16306 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
16307 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
16308 (*div<SVE_F>4): Likewise.
16310 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
16312 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
16315 2018-07-02 Richard Biener <rguenther@suse.de>
16317 PR tree-optimization/86363
16318 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
16319 memset argument refers to a non-variable address.
16321 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
16323 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
16324 {PLUS,MINUS}_EXPR code to...
16325 (adjust_symbolic_bound): ...here,
16326 (combine_bound): ...here,
16327 (set_value_range_with_overflow): ...and here.
16329 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
16331 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
16333 (extract_range_from_abs_expr): ...here.
16335 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
16337 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
16338 -fno-omit-frame-pointer when not optimizing.
16340 2018-07-02 Martin Liska <mliska@suse.cz>
16343 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
16344 (propagate_nothrow): Likewise.
16346 2018-07-02 Martin Liska <mliska@suse.cz>
16349 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
16351 2018-07-02 David Malcolm <dmalcolm@redhat.com>
16353 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
16354 function in r262149, changing "loc" param from source_location to
16355 const dump_location_t &.
16356 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
16357 declaration, as above.
16359 2018-07-01 Paul Koning <ni1d@arrl.net>
16361 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
16362 -munit-asm, -mgnu-asm, -mdec-asm.
16363 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
16364 (pdp11_output_labelref): New.
16365 (pdp11_output_def): New.
16366 (pdp11_output_addr_vec_elt): New.
16367 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
16368 %# and %@ format codes.
16369 (pdp11_option_override): New.
16370 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
16371 (pdp11_output_ident): New.
16372 (pdp11_asm_named_section): New.
16373 (pdp11_asm_init_sections): New.
16374 (pdp11_file_start): New.
16375 (pdp11_file_end): New.
16376 (output_ascii): Use .ascii/.asciz for -mdec-asm.
16377 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
16378 %o, like %c but octal.
16379 (pdp11_option_override): New.
16380 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
16382 (DATA_SECTION_ASM_OP): Ditto.
16383 (READONLY_DATA_SECTION_ASM_OP): New.
16384 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
16385 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
16386 (ASM_OUTPUT_LABELREF): Ditto.
16387 (ASM_OUTPUT_DEF): Ditto.
16388 (ASM_OUTPUT_EXTERNAL): New.
16389 (ASM_OUTPUT_SOURCE_FILENAME): New.
16390 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
16391 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
16392 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
16393 %# and %@ format codes.
16394 * config/pdp11/pdp11.opt (mgnu-asm): New.
16395 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
16396 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
16397 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
16399 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
16401 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
16402 dereferencing path[] beyond its length.
16404 (debug_all_paths): New.
16405 (rewire_first_differing_edge): New.
16406 (adjust_paths_after_duplication): New.
16407 (duplicate_thread_path): Call adjust_paths_after_duplication.
16409 (thread_through_all_blocks): Add new argument to
16410 duplicate_thread_path.
16412 2018-06-30 Jim Wilson <jimw@sifive.com>
16414 * config/riscv/predicates.md (p2m1_shift_operand): New.
16415 (high_mask_shift_operand): New.
16416 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
16417 pattern using p2m1_shift_operand.
16418 (lshsi3_zero_extend_3+2): New combiner pattern using
16419 high_mask_shift_operand.
16421 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16423 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
16425 (vect_recog_rotate_pattern): ...here.
16426 (vect_convert_input): Try to insert casts of invariants in the
16428 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
16429 preheader to be empty.
16431 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16433 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
16434 vector type. If given, install it in the new statement's
16435 STMT_VINFO_VECTYPE.
16436 (vect_element_precision): New function.
16437 (vect_unpromoted_value): New struct.
16438 (vect_unpromoted_value::vect_unpromoted_value): New function.
16439 (vect_unpromoted_value::set_op): Likewise.
16440 (vect_look_through_possible_promotion): Likewise.
16441 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
16442 (vect_widened_op_tree, vect_convert_input): Likewise.
16443 (vect_convert_inputs, vect_convert_output): Likewise.
16444 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
16445 to handle the optional cast of the multiplication result and
16446 vect_widened_op_tree to detect the widened multiplication itself.
16447 Do not require the input and output of promotion casts to have
16448 the same sign, but base the signedness of the operation on the
16449 input rather than the result. If the pattern includes two
16450 promotions, check that those promotions have the same sign.
16451 Do not restrict the MULT_EXPR handling to a double-width result;
16452 handle quadruple-width results and wider. Use vect_convert_inputs
16453 to convert the inputs to the common type.
16454 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
16455 to handle the optional cast of the ABS result. Also allow a sign
16456 change or a sign extension between the ABS and MINUS.
16457 Use vect_widened_op_tree to detect the widened subtraction and use
16458 vect_convert_inputs to convert the inputs to the common type.
16459 (vect_handle_widen_op_by_const): Delete.
16460 (vect_recog_widen_op_pattern): New function.
16461 (vect_recog_widen_mult_pattern): Use it.
16462 (vect_recog_widen_shift_pattern): Likewise.
16463 (vect_recog_widen_sum_pattern): Use
16464 vect_look_through_possible_promotion to handle the promoted
16467 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16469 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
16470 the containing gimple_seq *.
16471 * gimple-iterator.h (gsi_for_stmt): Declare it.
16472 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
16473 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
16474 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
16475 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
16476 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
16478 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
16480 (vect_mark_pattern_stmts): ...here. Handle cases in which the
16481 statement being replaced is part of an existing pattern
16482 definition sequence, inserting the new pattern statements before
16484 (vect_pattern_recog_1): Don't return a bool. If the statement
16485 is already part of a pattern, instead apply pattern matching
16486 to the pattern definition statements. Don't clear the
16487 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
16488 (vect_pattern_recog): Don't break after the first match;
16489 continue processing the pattern definition statements instead.
16490 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
16492 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16494 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
16495 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
16496 (vect_recog_widen_sum_pattern): Use it.
16498 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16500 * tree-vect-loop.c (vectorizable_reduction): Assert that the
16501 phi is not a pattern statement and has not been replaced by
16502 a pattern statement.
16503 * tree-vect-patterns.c (type_conversion_p): Don't check
16504 STMT_VINFO_IN_PATTERN_P.
16505 (vect_recog_vector_vector_shift_pattern): Likewise.
16506 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
16507 the pattern statement rather than the original statement; check
16508 directly for a WIDEN_MULT_EXPR here.
16509 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
16510 vect_is_simple_use to return the pattern statement rather
16511 than the original statement; use is_pattern_stmt_p to check
16512 for such a pattern statement.
16513 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
16514 to return the pattern statement rather than the original statement;
16515 don't do the same transformation here.
16516 (vect_is_simple_use): If the defining statement has been replaced
16517 by a pattern statement, return the pattern statement instead.
16518 Remove the corresponding (local) transformation from the vectype
16521 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16523 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
16524 end and default to null.
16525 * tree-vect-loop.c (vect_create_epilog_for_reduction)
16526 (vectorizable_reduction): Update calls accordingly, dropping the
16527 gimple ** argument if the passed-back statement isn't needed.
16528 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
16529 (vect_recog_rotate_pattern): Likewise.
16530 (vect_recog_mask_conversion_pattern): Likewise.
16531 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
16532 (vect_mask_constant_operand_p): Likewise.
16533 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
16534 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
16535 (get_group_load_store_type, get_load_store_type): Likewise.
16536 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
16537 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
16538 (vectorizable_conversion, vectorizable_assignment): Likewise.
16539 (vectorizable_shift, vectorizable_operation): Likewise.
16540 (vectorizable_store, vect_is_simple_cond): Likewise.
16541 (vectorizable_condition, vectorizable_comparison): Likewise.
16542 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
16543 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
16544 and move it to the end. Cope with null def_stmt_outs.
16546 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
16548 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
16550 2018-06-29 Jeff Law <law@redhat.com>
16552 * config/v850/v850.c (v850_legitimate_address_p): Handle large
16553 displacements for TARGET_V850E2V3 and newer.
16554 (TARGET_LRA_P): Remove. Defaults to LRA now.
16555 * config/v850/v850.md (sign23byte_load): Remove.
16556 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
16557 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
16559 2018-06-29 Martin Liska <mliska@suse.cz>
16562 * coverage.c (coverage_init): Mangle full path name.
16563 * doc/invoke.texi: Document the change.
16564 * gcov-io.c (mangle_path): New.
16565 * gcov-io.h (mangle_path): Likewise.
16566 * gcov.c (mangle_name): Use mangle_path for path mangling.
16568 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16570 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
16571 if starting source register is not even.
16573 2018-06-29 Martin Liska <mliska@suse.cz>
16575 PR tree-optimization/86263
16576 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
16577 Make edge redirection.
16579 2018-06-29 David Malcolm <dmalcolm@redhat.com>
16581 * dumpfile.c (dump_loc): Add indentation based on scope depth.
16582 (dump_scope_depth): New variable.
16583 (get_dump_scope_depth): New function.
16584 (dump_begin_scope): New function.
16585 (dump_end_scope): New function.
16586 * dumpfile.h (get_dump_scope_depth): New declaration.
16587 (dump_begin_scope): New declaration.
16588 (dump_end_scope): New declaration.
16589 (class auto_dump_scope): New class.
16590 (AUTO_DUMP_SCOPE): New macro.
16591 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
16594 2018-06-29 Richard Biener <rguenther@suse.de>
16596 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
16597 compute_all_dependences succeeds.
16598 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
16599 exceed --param loop-max-datarefs-for-datadeps.
16601 2018-06-29 Jakub Jelinek <jakub@redhat.com>
16603 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
16605 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
16608 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
16609 alternative 0 in preferred_for_speed attribute.
16611 2018-06-28 Paul Koning <ni1d@arrl.net>
16613 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
16614 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
16615 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
16616 * config/pdp11/pdp11.md: Correct "length" attribute calculation
16617 for shift insn patterns.
16619 2018-06-28 David Malcolm <dmalcolm@redhat.com>
16621 * cgraph.c (cgraph_node::get_body): Replace assignments to
16622 "dump_file" with calls to set_dump_file.
16623 * dumpfile.c (alt_dump_file): Make static, and group with...
16624 (alt_flags): ...this definition.
16625 (dumps_are_enabled): New variable.
16626 (refresh_dumps_are_enabled): New function.
16627 (set_dump_file): New function.
16628 (set_alt_dump_file): New function.
16629 (gcc::dump_manager::dump_start): Replace assignments to
16630 "dump_file" and "alt_dump_file" with calls to set_dump_file and
16632 (gcc::dump_manager::dump_finish): Likewise.
16633 * dumpfile.h (alt_dump_file): Delete decl.
16634 (dumps_are_enabled): New variable decl.
16635 (set_dump_file): New function decl.
16636 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
16638 * tree-nested.c (lower_nested_functions): Replace assignments to
16639 "dump_file" with calls to set_dump_file.
16641 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
16643 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
16644 goto_locus of each outgoing edge of each basic block.
16646 2018-06-28 Richard Biener <rguenther@suse.de>
16648 * dwarf2out.c (decl_scope_table): Remove.
16649 (push_decl_scope): Likewise.
16650 (pop_decl_scope): Likewise.
16651 (gen_type_die_for_member): Do not call push/pop_decl_scope.
16652 (gen_struct_or_union_type_die): Likewise.
16653 (gen_tagged_type_die): Likewise.
16654 (dwarf2out_init): Do not initialize decl_scope_table.
16655 (dwarf2out_c_finalize): Do not free it.
16657 2018-06-28 Richard Biener <rguenther@suse.de>
16659 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
16660 deciding whether to not re-use a DIE.
16662 2018-06-28 Richard Biener <rguenther@suse.de>
16664 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
16665 DW_AT_abstract_origin attribute.
16667 2018-06-28 Martin Liska <mliska@suse.cz>
16669 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
16670 Use newly introduced constants.
16671 * tree-switch-conversion.h (struct jump_table_cluster):
16672 Define max_ratio_for_size and max_ratio_for_speed.
16674 2018-06-28 Martin Liska <mliska@suse.cz>
16676 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
16677 Add new checking assert to catch invalid state.
16678 (jump_table_cluster::can_be_handled): Handle single case
16680 (jump_table_cluster::is_beneficial): Bail out for such case.
16681 (bit_test_cluster::find_bit_tests):
16682 Add new checking assert to catch invalid state.
16683 (bit_test_cluster::can_be_handled): Handle single case
16685 (bit_test_cluster::is_beneficial): Bail out for such case.
16686 (switch_decision_tree::analyze_switch_statement):
16689 2018-06-28 Martin Liska <mliska@suse.cz>
16691 * common.opt: Introduce -completion option.
16692 * gcc.c (driver_handle_option): Handle it.
16693 (driver::main): Print completions if completion
16695 * opt-suggestions.c (option_proposer::get_completions):
16697 (option_proposer::suggest_completion): Likewise.
16698 (option_proposer::find_param_completions): Likewise.
16699 (verify_autocompletions): Likewise.
16700 (test_completion_valid_options): Likewise.
16701 (test_completion_valid_params): Likewise.
16702 (in_completion_p): Likewise.
16703 (empty_completion_p): Likewise.
16704 (test_completion_partial_match): Likewise.
16705 (test_completion_garbage): Likewise.
16706 (opt_proposer_c_tests): Likewise.
16707 * opt-suggestions.h: Declare new functions.
16708 * opts.c (common_handle_option): Handle OPT__completion_.
16709 * selftest-run-tests.c (selftest::run_tests): Add
16710 opt_proposer_c_tests.
16711 * selftest.c (assert_str_startswith): New.
16712 * selftest.h (assert_str_startswith): Likewise.
16713 (opt_proposer_c_tests): New.
16714 (ASSERT_STR_STARTSWITH): Likewise.
16716 2018-06-28 Martin Liska <mliska@suse.cz>
16718 * Makefile.in: Add opt-suggestions.o.
16719 * gcc-main.c: Include opt-suggestions.h.
16720 * gcc.c (driver::driver): Likewise.
16721 (driver::~driver): Remove m_option_suggestions.
16722 (driver::build_option_suggestions): Moved to option_proposer.
16723 (driver::suggest_option): Likewise.
16724 (driver::handle_unrecognized_options): Use option_proposer.
16725 * gcc.h (class driver): Add new memver m_option_proposer.
16726 * opt-suggestions.c: New file.
16727 * opt-suggestions.h: New file.
16729 2018-06-28 Martin Liska <mliska@suse.cz>
16731 * vec.h (class auto_string_vec): New (moved from auto_argvec).
16732 (auto_string_vec::~auto_string_vec): Likewise.
16734 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
16736 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
16737 prevent_decl_creation_for_types fields up and add reset_location field.
16738 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
16739 statement if id->reset_location is true.
16740 (copy_edges_for_bb): Do not set goto_locus on the new edges if
16741 id->reset_location is true.
16742 (copy_phis_for_bb): Force input_location on the arguments if
16743 id->reset_location is true.
16744 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
16745 is set on the function to be inlined.
16747 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
16749 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
16751 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
16753 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
16754 registers for Pmode.
16755 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
16756 hard registers for the clobbered pseudo.
16758 2018-06-27 Paul Koning <ni1d@arrl.net>
16760 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
16761 mutually exclusive options.
16762 * config/pdp11/constraints.md (h): New constraint.
16763 (O): Update definition to match shift code generation.
16764 (D): New constraint.
16765 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
16767 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
16769 (output_jump): Change arguments.
16770 (pdp11_fixed_cc_regs): New function.
16771 (pdp11_cc_mode): Ditto.
16772 (pdp11_expand_shift): Ditto.
16773 (pdp11_assemble_shift): Ditto.
16774 (pdp11_small_shift): Ditto.
16775 (pdp11_branch_cost): Remove.
16776 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
16778 (pdp11_register_move_cost): Update for CC registers.
16779 (pdp11_rtx_costs): Add case for LSHIFTRT.
16780 (pdp11_output_jump): Add CCNZ mode conditional branches.
16781 (notice_update_cc_on_set): Remove.
16782 (pdp11_cc_mode): New function.
16783 (simple_memory_operand): Correct pre/post decrement case.
16784 (no_side_effect_operand): New function.
16785 (pdp11_regno_reg_class): Add CC_REGS class.
16786 (pdp11_fixed_cc_regs): New function.
16787 (pdp11_small_shift): New function.
16788 (pdp11_expand_shift): New function to expand shift insns.
16789 (pdp11_assemble_shift): New function to output shifts.
16790 (pdp11_branch_cost): Remove.
16791 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
16792 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
16793 (WCHAR_TYPE): Ditto.
16794 (PTRDIFF_TYPE): Ditto.
16795 (ADJUST_INSN_LENGTH): New macro.
16796 (FIXED_REGISTERS): Add CC registers.
16797 (CALL_USED_REGISTERS): Ditto.
16798 (reg_class): Ditto.
16799 (REG_CLASS_NAMES): Ditto.
16800 (REG_CLASS_CONTENTS): Ditto.
16801 (SELECT_CC_MODE): Use new function.
16802 (TARGET_FLAGS_REGNUM): New macro.
16803 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
16804 (cc0_reg_rtx): Remove.
16805 (CC_STATUS_MDEP): Remove.
16806 (CC_STATUS_MDEFP_INIT): Remove.
16807 (CC_IN_FPU): Remove.
16808 (NOTICE_UPDATE_CC): Remove.
16809 (REGISTER_NAMES): Add CC registers.
16810 (BRANCH_COST): Change to constant 1.
16811 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
16813 * config/pdp11/pdp11.opt (mbcopy): Remove.
16814 (mbcopy-builtin): Remove.
16815 (mbranch-cheap): Remove.
16816 (mbranch-expensive): Remove.
16817 * config/pdp11/predicates.md (expand_shift_operand): Update to
16818 match shift code generation.
16819 (ccnz_operator): New predicate.
16820 * doc/invoke.texi (PDP-11 Options): Remove deleted options
16821 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
16822 Remove non-existent option -mabshi, -mno-abshi. Document mutually
16824 * doc/md.texi (PDP-11): Document new D and h constraints. Update
16825 description of O constraint.
16827 2018-06-27 Jeff Law <law@redhat.com>
16828 Austin Law <austinklaw@gmail.com>
16830 * config/v850/v850.md (addsi3_set_flags): New pattern.
16831 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
16832 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
16833 (zero_extendhisi2_v850_set_flags): Likewise.
16834 (zero_extendqisi2_v850_set_flags): Likewise.
16835 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
16836 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
16837 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
16839 * config/v850/v850-protos.h (notice_update_cc): Remove.
16840 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
16841 (v850_print_operand): Handle 'D' and "d".
16842 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
16843 Add handling of arithmetic/logical operations compared against zero.
16844 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
16845 Do not look at v850_compare_op, instead get mode from last argument.
16846 (v850_gen_compare): Remove
16847 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
16848 after reload for prologue insns.
16849 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
16851 (construct_save_jarl): Likewise.
16852 (TARGET_FLAGS_REGNUM): Define.
16853 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
16854 (NOTICE_UPDATE_CC): Remove.
16855 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
16856 than cc0. Conditionalize on reload_completed.
16857 (cmpsi_insn, setfcc_insn): Likewise.
16858 (tst1 splitter): Turn into define_and_split which sets the flags
16860 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
16861 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
16862 (cstoresf4, cstoredf4): Clobber the flags.
16863 (cmpsi, cmpsf, cmpdf): Remove expanders.
16864 (setf_insn): Remove pattern.
16865 (addsi3): Turn into define_and_split which clobbers the flags after
16866 reload and a suitable pattern (addsi3_clobber_flags) for use after
16868 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
16869 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
16870 (ashrsi3, ashrsi3_v850e2): Likewise.
16871 (bins): Clobber the flags.
16872 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
16873 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
16874 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
16875 (call_value_internal_short, call_value_internal_long): Likewise.
16876 (callt_save_interrupt, callt_return_interrupt): Likewise.
16877 (save_interrupt, return_interrupt): Likewise.
16878 (callt_save_all_interrupt, save_all_interrupt): Likewise.
16879 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
16880 (restore_all_interrupt, _restore_all_interrupt): Likewise.
16881 (All FP comparisons): Only allow after reload has completed.
16883 (divh, divhu): Tweak output template.
16884 (branch_z_normal, branch_z_invert): Remove
16885 (branch_nz_normal, branch_nz_invert): Likewise.
16886 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
16888 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
16889 * config/v850/v850.c (notice_update_cc): Remove.
16890 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
16891 (CC_NO_CARRY): Likewise.
16892 (NOTICE_UPDATE_CC): Define to nothing.
16893 * config/v850/v850.md: Remove block comment on cc0 handling
16894 Remove "cc" attribute from all patterns. Remove cc_status handling
16895 from all patterns. Minor formatting fixes.
16897 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16899 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
16900 (cortex-a76.cortex-a55): Likewise.
16901 * config/aarch64/aarch64-tune.md: Regenerate.
16902 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
16903 cortex-a76.cortex-a55.
16905 2018-06-27 Jeff Law <law@redhat.com>
16907 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
16908 (MULTILIB_DIRNAMES): Similarly.
16910 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
16912 * gimple.h (gimple_return_retbnd): Delete.
16913 (gimple_return_set_retbnd): Likewise.
16914 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
16915 gimple_return_set_retbnd.
16916 * gimple-pretty-print.c (dump_gimple_return): Remove call to
16917 gimple_return_retbnd and adjust.
16918 * tree-inline.h (struct copy_body_data): Remove retbnd field.
16919 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
16920 Explicitly return NULL in a couple more cases. Move assertion
16921 on debug statements and remove unreachable code.
16922 (reset_debug_binding): Do not test id->retbnd.
16923 (expand_call_inline): Do not set it.
16925 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
16927 * configure.ac: Add --disable-gcov option.
16928 * configure: Regenerate.
16929 * Makefile.in: Honour @enable_gcov@.
16930 * doc/install.texi: Document --disable-gcov.
16932 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16934 * config/arm/arm-cpus.in (cortex-a76): New entry.
16935 (cortex-a76.cortex-a55): Likewise.
16936 * config/arm/arm-tables.opt: Regenerate.
16937 * config/arm/arm-tune.md: Likewise.
16938 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
16939 * doc/invoke.texi (ARM Options): Document cortex-a76 and
16940 cortex-a76.cortex-a55.
16942 2018-06-27 Tamar Christina <tamar.christina@arm.com>
16945 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
16947 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
16949 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
16951 (EPILOGUE_USES): Likewise.
16953 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
16955 * tree-inline.c (remap_location): New function extracted from...
16956 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
16957 (copy_phis_for_bb): ...here. Call remap_location.
16958 (copy_cfg_body): Adjust call to copy_edges_for_bb.
16960 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
16962 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
16963 unaligned vsx for 16B memset.
16965 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
16968 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
16969 ieee128_float_type_node to long_double_type_node unless
16970 TARGET_LONG_DOUBLE_128 is set.
16972 2018-06-26 David Malcolm <dmalcolm@redhat.com>
16974 * cfgloop.c (get_loop_location): Convert return type from
16975 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
16976 by implicit construction from rtx_insn *, and using
16977 dump_user_location_t::from_function_decl for the fallback case.
16978 * cfgloop.h (get_loop_location): Convert return type from
16979 location_t to dump_user_location_t.
16980 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
16981 dump_printf_loc to pass in a dump_location_t rather than a
16982 location_t, via the gimple stmt.
16983 * coverage.c (get_coverage_counts): Update calls to
16984 dump_printf_loc to pass in dump_location_t rather than a
16986 * doc/optinfo.texi (Dump types): Convert example of
16987 dump_printf_loc from taking "locus" to taking "insn". Update
16988 description of the "_loc" calls to cover dump_location_t.
16989 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
16991 (dump_user_location_t::dump_user_location_t): New constructors,
16992 from gimple *stmt and rtx_insn *.
16993 (dump_user_location_t::from_function_decl): New function.
16994 (dump_loc): Make static.
16995 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
16996 const dump_location_t &.
16997 (dump_generic_expr_loc): Delete.
16998 (dump_printf_loc): Convert param "loc" from location_t to
16999 const dump_location_t &.
17000 (selftest::test_impl_location): New function.
17001 (selftest::dumpfile_c_tests): New function.
17002 * dumpfile.h: Include "profile-count.h".
17003 (class dump_user_location_t): New class.
17004 (struct dump_impl_location_t): New struct.
17005 (class dump_location_t): New class.
17006 (dump_printf_loc): Convert 2nd param from source_location to
17007 const dump_location_t &.
17008 (dump_generic_expr_loc): Delete.
17009 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
17010 const dump_location_t &.
17011 * gimple-fold.c (fold_gimple_assign): Update call to
17012 dump_printf_loc to pass in a dump_location_t rather than a
17013 location_t, via the gimple stmt.
17014 (gimple_fold_call): Likewise.
17015 * gimple-loop-interchange.cc
17016 (loop_cand::analyze_iloop_reduction_var): Update for change to
17017 check_reduction_path.
17018 (tree_loop_interchange::interchange): Update for change to
17019 find_loop_location.
17020 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
17021 change in return-type of find_loop_location.
17022 (graphite_regenerate_ast_isl): Likewise.
17023 * graphite-optimize-isl.c (optimize_isl): Likewise.
17024 * graphite.c (graphite_transform_loops): Likewise.
17025 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
17026 pass in a dump_location_t rather than a location_t, via the
17028 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
17029 * ipa.c (walk_polymorphic_call_targets): Likewise.
17030 * loop-unroll.c (report_unroll): Convert "locus" param from
17031 location_t to dump_location_t.
17032 (decide_unrolling): Update for change to get_loop_location's
17034 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
17035 location_t to dump_user_location_t.
17036 (grid_find_single_omp_among_assignments_1): Updates calls to
17037 dump_printf_loc to pass in a dump_location_t rather than a
17038 location_t, via the gimple stmt.
17039 (grid_parallel_clauses_gridifiable): Convert "tloc" from
17040 location_t to dump_location_t. Updates calls to dump_printf_loc
17041 to pass in a dump_location_t rather than a location_t, via the
17043 (grid_inner_loop_gridifiable_p): Likewise.
17044 (grid_dist_follows_simple_pattern): Likewise.
17045 (grid_gfor_follows_tiling_pattern): Likewise.
17046 (grid_target_follows_gridifiable_pattern): Likewise.
17047 (grid_attempt_target_gridification): Convert initialization
17048 of local "grid" from memset to zero-initialization; FIXME: does
17049 this require C++11? Update call to dump_printf_loc to pass in a
17050 optinfo_location rather than a location_t, via the gimple stmt.
17051 * profile.c (read_profile_edge_counts): Updates call to
17052 dump_printf_loc to pass in a dump_location_t rather than a
17054 (compute_branch_probabilities): Likewise.
17055 * selftest-run-tests.c (selftest::run_tests): Call
17057 * selftest.h (dumpfile_c_tests): New decl.
17058 * tree-loop-distribution.c (pass_loop_distribution::execute):
17059 Update for change in return type of find_loop_location.
17060 * tree-parloops.c (parallelize_loops): Likewise.
17061 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
17062 "locus" from location_t to dump_user_location_t.
17063 (canonicalize_loop_induction_variables): Likewise.
17064 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
17065 for change in return type of find_loop_location.
17066 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
17067 to dump_printf_loc to pass in a dump_location_t rather than a
17068 location_t, via the stmt.
17069 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
17071 * tree-vect-loop-manip.c (find_loop_location): Convert return
17072 type from source_location to dump_user_location_t.
17073 (vect_do_peeling): Update for above change.
17074 (vect_loop_versioning): Update for change in type of
17076 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
17077 from location_t to dump_user_location_t.
17078 (vect_estimate_min_profitable_iters): Update for change in type
17080 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
17081 location_t to dump_location_t.
17082 (vect_slp_bb): Update for change in type of vect_location.
17083 * tree-vectorizer.c (vect_location): Convert from source_location
17084 to dump_user_location_t.
17085 (try_vectorize_loop_1): Update for change in vect_location's type.
17086 (vectorize_loops): Likewise.
17087 (increase_alignment): Likewise.
17088 * tree-vectorizer.h (vect_location): Convert from source_location
17089 to dump_user_location_t.
17090 (find_loop_location): Convert return type from source_location to
17091 dump_user_location_t.
17092 (check_reduction_path): Convert 1st param from location_t to
17093 dump_user_location_t.
17094 * value-prof.c (check_counter): Update call to dump_printf_loc to
17095 pass in a dump_user_location_t rather than a location_t; update
17096 call to error_at for change in type of "locus".
17097 (check_ic_target): Update call to dump_printf_loc to
17098 pass in a dump_user_location_t rather than a location_t, via the
17101 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
17103 * config/s390/s390.h (enum processor_flags): Do not use
17104 default tune parameter when -march was specified.
17106 2018-06-26 Jakub Jelinek <jakub@redhat.com>
17109 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
17110 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
17113 2018-06-26 Richard Biener <rguenther@suse.de>
17115 PR tree-optimization/86287
17117 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
17118 (vect_analyze_loop): Initialize n_stmts.
17120 2018-06-26 Richard Biener <rguenther@suse.de>
17122 PR middle-end/86271
17123 * fold-const.c (fold_convertible_p): Pointer extension
17126 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
17129 * dwarf2out.c (loc_list_has_views): Adjust comments.
17130 (dw_loc_list): Split single cross-partition range with
17133 2018-06-25 Jeff Law <law@redhat.com>
17135 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
17136 on -mbig-switch by default.
17138 * config/v850/predicates.md (const_float_1_operand): Fix match_code
17140 (const_float_0_operand): Remove unused predicate.
17141 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
17142 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
17143 (recipsf2): New expander. Original pattern now called
17145 (recipdf2, recipdf2_insn): Similarly.
17146 (rsqrtsf2, rsqrtsf2_insn): Similarly
17147 (rsqrtdf2, rsqrtdf2_insn): Similarly
17149 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
17151 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
17152 Simplify logic for FreeBSD (twice).
17154 2018-06-25 Martin Sebor <msebor@redhat.com>
17156 PR tree-optimization/86204
17157 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
17158 a strnlen result if it's less than the length of the string.
17160 2018-06-25 Martin Sebor <msebor@redhat.com>
17162 PR tree-optimization/85700
17163 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
17164 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
17165 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
17167 2018-06-25 Martin Sebor <msebor@redhat.com>
17169 * doc/extend.texi (Zero-length arrays): Update and clarify.
17171 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
17173 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
17174 added IEEE/IBM long double multilib support on PowerPC little
17175 endian Linux systems.
17176 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
17177 (MULTILIB_DEFAULTS): Likewise.
17178 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17180 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
17181 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
17182 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
17184 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
17186 PR middle-end/86311
17187 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
17188 (REORDER_45): Likewise.
17190 2018-06-25 Jeff Law <law@redhat.com>
17192 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
17193 dividend to 32 bits. Adjust length.
17194 (udivmodhi4): Cleanup output template. Fix length.
17196 2018-06-25 Carl Love <cel@us.ibm.com>
17198 * config/rs6000/vsx.md: Change word selector to prefered location.
17200 2018-06-25 Richard Biener <rguenther@suse.de>
17202 PR tree-optimization/86304
17203 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
17204 epilogue-if-converted loops as well.
17206 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
17208 * lto-section-out.c (lto_begin_section): Do not print section
17209 name for noaddr and unnumbered dumps.
17211 2018-06-25 Richard Biener <rguenther@suse.de>
17213 * tree-vectorizer.h (struct vec_info_shared): New structure
17214 with parts split out from struct vec_info and loop_nest from
17215 struct _loop_vec_info.
17216 (struct vec_info): Adjust accordingly.
17217 (struct _loop_vec_info): Likewise.
17218 (LOOP_VINFO_LOOP_NEST): Adjust.
17219 (LOOP_VINFO_DATAREFS): Likewise.
17220 (LOOP_VINFO_DDRS): Likewise.
17221 (struct _bb_vec_info): Likewise.
17222 (BB_VINFO_DATAREFS): Likewise.
17223 (BB_VINFO_DDRS): Likewise.
17224 (struct _stmt_vec_info): Add dr_aux member.
17225 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
17226 (DR_MISALIGNMENT_UNINITIALIZED): New.
17227 (set_dr_misalignment): Adjust.
17228 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
17229 (vect_analyze_loop): Adjust prototype.
17230 (vect_analyze_loop_form): Likewise.
17231 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
17232 Compute dependences lazily.
17233 (vect_record_base_alignments): Use shared datarefs/ddrs.
17234 (vect_verify_datarefs_alignment): Likewise.
17235 (vect_analyze_data_refs_alignment): Likewise.
17236 (vect_analyze_data_ref_accesses): Likewise.
17237 (vect_analyze_data_refs): Likewise.
17238 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
17239 constructor parameter for shared part.
17240 (vect_analyze_loop_form): Pass in shared part and adjust.
17241 (vect_analyze_loop_2): Pass in storage for the number of
17242 stmts. Move loop nest finding to the caller. Compute
17244 (vect_analyze_loop): Pass in shared part.
17245 (vect_transform_loop): Verify shared datarefs are unchanged.
17246 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
17247 constructor parameter for shared part.
17248 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
17249 (vect_slp_bb): Verify shared datarefs are unchanged before
17251 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
17253 (new_stmt_vec_info): Initialize DR_AUX misalignment to
17254 DR_MISALIGNMENT_UNINITIALIZED.
17255 * tree-vectorizer.c (vec_info::vec_info): Add constructor
17256 parameter for shared part.
17257 (vec_info::~vec_info): Adjust.
17258 (vec_info_shared::vec_info_shared): New.
17259 (vec_info_shared::~vec_info_shared): Likewise.
17260 (vec_info_shared::save_datarefs): Likewise.
17261 (vec_info_shared::check_datarefs): Likewise.
17262 (try_vectorize_loop_1): Construct shared part live for analyses
17263 of a single loop for multiple vector sizes.
17264 * tree-parloops.c (gather_scalar_reductions): Adjust.
17266 2018-06-25 Richard Biener <rguenther@suse.de>
17268 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
17269 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
17270 (vect_analyze_data_refs): Remove similar code from here and
17271 simplify accordingly.
17273 2018-06-25 Richard Biener <rguenther@suse.de>
17275 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
17276 for reverse storage order accesses rather than asserting
17277 they cannot happen here.
17279 2018-06-25 Tom de Vries <tdevries@suse.de>
17282 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
17283 Use data16 instead of .byte for insn prefix.
17285 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
17288 * parser.c (make_char_string_pack): Pass this literal chars
17289 through cpp_interpret_string.
17290 (cp_parser_userdef_numeric_literal): Check the result of
17291 make_char_string_pack.
17293 2018-06-24 Maya Rashish <coypu@sdf.org>
17295 * ginclude/stddef.h: Simplify conditions around avoiding
17296 re-definition of __size_t.
17298 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
17300 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
17301 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
17303 2018-06-22 Maya Rashish <coypu@sdf.org>
17305 * doc/invoke.texi (mno-fancy-math-387): Update for changes
17306 made to OpenBSD and NetBSD through the years.
17308 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
17310 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
17311 behavior of vec_pack (vector double, vector double) to match
17312 behavior of vec_float2 (vector double, vector double).
17314 2018-06-22 Olivier Hainque <hainque@adacore.com>
17316 * gimplify.c (gimplify_function_tree): Prevent creation
17317 of a trampoline for the address of the current function
17318 passed to entry/exit instrumentation hooks.
17320 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
17323 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
17326 2018-06-22 Martin Liska <mliska@suse.cz>
17328 PR tree-optimization/86263
17329 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
17330 Bail out if is_enabled is false.
17331 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
17333 (jump_table_cluster::is_enabled): New function.
17335 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
17337 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
17338 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
17339 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
17340 (lto_input_ts_binfo_tree_pointers): Likewise.
17341 * tree-streamer-out.c (streamer_write_tree_bitfields,
17342 write_ts_binfo_tree_pointers): Likewise.
17343 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
17345 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
17347 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
17349 2018-06-22 Martin Liska <mliska@suse.cz>
17351 * symbol-summary.h (get): Make it pure and inline move
17352 functionality from ::get function.
17353 (get): Remove and inline into ::get and ::get_create.
17354 (get_create): Move code from ::get function.
17356 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17359 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
17360 -x assembler-with-cpp.
17362 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17364 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
17365 _FILE_OFFSET_BITS=64 for C++.
17367 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
17369 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
17370 conversion insn that shows up when pr85657-3.c is compiled using
17371 IEEE 128-bit long double.
17372 (neg<mode>2_internal): Use the correct mode to check whether the
17373 mode is IBM extended.
17374 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
17375 multiply and divide external functions from being created more
17378 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
17380 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
17382 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
17383 the edge can be forwarded.
17384 (cfg_layout_merge_blocks): Likewise.
17386 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
17388 * except.c (finish_eh_generation): Commit edge insertions only after
17389 the EH edges have been redirected from post-landing to landing pads.
17391 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
17393 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
17394 create_tmp_var_for to create the FRAME decl.
17395 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
17397 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
17399 * tree-inline.c (copy_edges_for_bb): Minor tweak.
17400 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
17401 debug statement when resetting its value.
17402 (expand_call_inline): Copy the locus of the call onto the assignment
17403 of the return value, if any. Use local variable in more cases.
17405 2018-06-21 Martin Liska <mliska@suse.cz>
17407 * ipa-pure-const.c (propagate_nothrow): Use
17408 funct_state_summaries->get.
17409 (dump_malloc_lattice): Likewise.
17410 (propagate_malloc): Likewise.
17412 2018-06-21 Richard Biener <rguenther@suse.de>
17414 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
17415 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
17416 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
17418 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
17419 BLOCK_ABSTRACT_ORIGIN unconditionally.
17421 2018-06-21 David Malcolm <dmalcolm@redhat.com>
17423 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
17425 * ipa-reference.c (ipa_reference_c_finalize): Delete
17426 ipa_ref_opt_sum_summaries and set it to NULL.
17428 2018-06-21 Tom de Vries <tdevries@suse.de>
17430 PR tree-optimization/85859
17431 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
17432 test with comment from bb_no_side_effects_p.
17434 2018-06-21 Richard Biener <rguenther@suse.de>
17436 PR tree-optimization/86232
17437 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
17438 max for constant niter.
17440 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
17442 * config/aarch64/aarch64-simd.md
17443 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
17445 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
17447 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
17448 Make opernads of the unspec commutative.
17450 2018-06-21 Richard Biener <rguenther@suse.de>
17452 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
17453 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
17454 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
17455 (vect_analyze_data_ref_dependence): Re-order checks to deal with
17457 (vect_record_base_alignments): Do not record base alignment
17458 for gathers or scatters.
17459 (vect_compute_data_ref_alignment): Drop return value that is always
17460 true. Bail out early for gathers or scatters.
17461 (vect_enhance_data_refs_alignment): Bail out early for gathers
17463 (vect_find_same_alignment_drs): Likewise.
17464 (vect_analyze_data_refs_alignment): Remove dead code.
17465 (vect_slp_analyze_and_verify_node_alignment): Likewise.
17466 (vect_analyze_data_refs): For possible gathers or scatters do
17467 not create an alternate DR, just check their possible validity
17468 and mark them. Adjust DECL_NONALIASED handling to not rely
17469 on DR_BASE_ADDRESS.
17470 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
17471 update inits of gathers or scatters.
17472 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
17473 Also copy gather/scatter flag to pattern vinfo.
17475 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
17477 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
17478 behavior of vec_packsu (vector unsigned long long, vector unsigned
17479 long long) to match behavior of vec_packs with same signature.
17481 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
17482 Thomas Schwinge <thomas@codesourcery.com>
17483 Cesar Philippidis <cesar@codesourcery.com>
17485 * gimplify.c (gimplify_scan_omp_clauses): Add support for
17486 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
17487 (gimplify_adjust_omp_clauses): Likewise.
17488 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
17489 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
17490 (gimplify_omp_target_update): Update handling of acc update and
17492 * omp-low.c (install_var_field): Remove unused parameter
17493 base_pointers_restrict.
17494 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
17495 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
17497 (omp_target_base_pointers_restrict_p): Delete.
17498 (scan_omp_target): Update call to scan_sharing_clauses.
17499 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
17501 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
17502 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
17503 (convert_local_omp_clauses): Likewise.
17504 * tree-pretty-print.c (dump_omp_clause): Likewise.
17505 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
17507 (omp_clause_code_name): Likewise.
17509 2018-06-20 Jakub Jelinek <jakub@redhat.com>
17512 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
17515 PR tree-optimization/86231
17516 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
17517 anti-range don't overwrite *vr0min before using it to compute *vr0max.
17519 2018-06-20 Tom de Vries <tdevries@suse.de>
17521 PR tree-optimization/86097
17522 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
17523 iv type if signedness of iv type is not the same as that of *nit.
17525 2018-06-20 Jakub Jelinek <jakub@redhat.com>
17527 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
17528 EDGE_EH edges, verify they are all EDGE_EH.
17530 2018-06-20 Maya Rashish <coypu@sdf.org>
17532 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
17534 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17536 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
17537 * config/aarch64/aarch64.c (xgene1_tunings): Add
17538 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
17539 (aarch64_mode_valid_for_sched_fusion_p):
17540 Allow 16-byte modes.
17541 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
17542 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
17544 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
17546 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
17547 * config/aarch64/iterators.md (VQ2): New mode iterator.
17549 2018-06-20 Martin Liska <mliska@suse.cz>
17551 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
17552 Change default ratio from 10 to 8.
17554 2018-06-20 Martin Liska <mliska@suse.cz>
17556 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
17558 (bit_test_cluster::find_bit_tests): Likewise.
17559 (switch_decision_tree::analyze_switch_statement): Find clusters.
17560 * tree-switch-conversion.h (struct jump_table_cluster): Document
17563 2018-06-20 Martin Liska <mliska@suse.cz>
17565 * tree-switch-conversion.c (switch_conversion::collect):
17566 Record m_uniq property.
17567 (switch_conversion::expand): Bail out for special conditions.
17568 (group_cluster::~group_cluster): New.
17569 (group_cluster::group_cluster): Likewise.
17570 (group_cluster::dump): Likewise.
17571 (jump_table_cluster::emit): New.
17572 (switch_decision_tree::fix_phi_operands_for_edges): New.
17573 (struct case_node): Remove struct.
17574 (jump_table_cluster::can_be_handled): New.
17575 (case_values_threshold): Moved to header.
17576 (reset_out_edges_aux): Likewise.
17577 (jump_table_cluster::is_beneficial): New.
17578 (bit_test_cluster::can_be_handled): Likewise.
17579 (add_case_node): Remove.
17580 (bit_test_cluster::is_beneficial): New.
17581 (case_bit_test::cmp): New.
17582 (bit_test_cluster::emit): New.
17583 (expand_switch_as_decision_tree_p): Remove.
17584 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
17585 (fix_phi_operands_for_edge): Likewise.
17586 (switch_decision_tree::analyze_switch_statement): New.
17587 (compute_cases_per_edge): Move ...
17588 (switch_decision_tree::compute_cases_per_edge): ... here.
17589 (try_switch_expansion): Likewise.
17590 (switch_decision_tree::try_switch_expansion): Likewise.
17591 (record_phi_operand_mapping): Likewise.
17592 (switch_decision_tree::record_phi_operand_mapping): Likewise.
17593 (emit_case_decision_tree): Likewise.
17594 (switch_decision_tree::emit): Likewise.
17595 (balance_case_nodes): Likewise.
17596 (switch_decision_tree::balance_case_nodes): Likewise.
17597 (dump_case_nodes): Likewise.
17598 (switch_decision_tree::dump_case_nodes): Likewise.
17599 (emit_jump): Likewise.
17600 (switch_decision_tree::emit_jump): Likewise.
17601 (emit_cmp_and_jump_insns): Likewise.
17602 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
17603 (emit_case_nodes): Likewise.
17604 (switch_decision_tree::emit_case_nodes): Likewise.
17605 (conditional_probability): Remove.
17606 * tree-switch-conversion.h (enum cluster_type): New.
17608 (struct cluster): Likewise.
17609 (cluster::cluster): Likewise.
17610 (struct simple_cluster): Likewise.
17611 (simple_cluster::simple_cluster): Likewise.
17612 (struct group_cluster): Likewise.
17613 (struct jump_table_cluster): Likewise.
17614 (struct bit_test_cluster): Likewise.
17615 (struct min_cluster_item): Likewise.
17616 (struct case_tree_node): Likewise.
17617 (case_tree_node::case_tree_node): Likewise.
17618 (jump_table_cluster::case_values_threshold): Likewise.
17619 (struct case_bit_test): Likewise.
17620 (struct switch_decision_tree): Likewise.
17621 (struct switch_conversion): Likewise.
17622 (switch_decision_tree::reset_out_edges_aux): Likewise.
17624 2018-06-20 Martin Liska <mliska@suse.cz>
17626 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
17627 (hoist_edge_and_branch_if_true): Likewise.
17628 (expand_switch_using_bit_tests_p): Likewise.
17629 (struct case_bit_test): Likewise.
17630 (case_bit_test_cmp): Likewise.
17631 (emit_case_bit_tests): Likewise.
17632 (switch_conversion::switch_conversion): New class.
17633 (struct switch_conv_info): Remove old struct.
17634 (collect_switch_conv_info): More to ...
17635 (switch_conversion::collect): ... this.
17636 (check_range): Likewise.
17637 (switch_conversion::check_range): Likewise.
17638 (check_all_empty_except_final): Likewise.
17639 (switch_conversion::check_all_empty_except_final): Likewise.
17640 (check_final_bb): Likewise.
17641 (switch_conversion::check_final_bb): Likewise.
17642 (create_temp_arrays): Likewise.
17643 (switch_conversion::create_temp_arrays): Likewise.
17644 (free_temp_arrays): Likewise.
17645 (gather_default_values): Likewise.
17646 (switch_conversion::gather_default_values): Likewise.
17647 (build_constructors): Likewise.
17648 (switch_conversion::build_constructors): Likewise.
17649 (constructor_contains_same_values_p): Likewise.
17650 (switch_conversion::contains_same_values_p): Likewise.
17651 (array_value_type): Likewise.
17652 (switch_conversion::array_value_type): Likewise.
17653 (build_one_array): Likewise.
17654 (switch_conversion::build_one_array): Likewise.
17655 (build_arrays): Likewise.
17656 (switch_conversion::build_arrays): Likewise.
17657 (gen_def_assigns): Likewise.
17658 (switch_conversion::gen_def_assigns): Likewise.
17659 (prune_bbs): Likewise.
17660 (switch_conversion::prune_bbs): Likewise.
17661 (fix_phi_nodes): Likewise.
17662 (switch_conversion::fix_phi_nodes): Likewise.
17663 (gen_inbound_check): Likewise.
17664 (switch_conversion::gen_inbound_check): Likewise.
17665 (process_switch): Use the newly created class.
17666 (switch_conversion::expand): New.
17667 (switch_conversion::~switch_conversion): New.
17668 * tree-switch-conversion.h: New file.
17670 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17672 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
17673 tree-vect-patterns.c.
17674 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
17675 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
17676 (vect_recog_sad_pattern): Likewise.
17677 (vect_recog_widen_sum_pattern): Likewise.
17678 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
17679 (vect_recog_widen_shift_pattern): Remove the type_in argument.
17680 (vect_recog_rotate_pattern): Likewise.
17681 (vect_recog_mult_pattern): Likewise.
17682 (vect_recog_vector_vector_shift_pattern): Likewise.
17683 (vect_recog_divmod_pattern): Likewise.
17684 (vect_recog_mixed_size_cond_pattern): Likewise.
17685 (vect_recog_bool_pattern): Likewise.
17686 (vect_recog_mask_conversion_pattern): Likewise.
17687 (vect_try_gather_scatter_pattern): Likewise.
17688 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
17689 (vect_recog_over_widening_pattern): Likewise.
17690 (vect_recog_gather_scatter_pattern): Likewise.
17691 (vect_recog_func_ptr): Move from tree-vectorizer.h
17692 (vect_vect_recog_func_ptrs): Move further down the file.
17693 (vect_recog_func): Likewise. Remove the third argument.
17694 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
17695 (vect_pattern_recog_1): Expect the pattern function to do any
17696 necessary target tests. Also expect it to provide a vector type.
17697 Remove the type_in handling.
17699 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17701 * tree-vect-patterns.c (vect_pattern_detected): New function.
17702 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
17703 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
17704 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
17705 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
17706 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
17707 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
17708 (vect_recog_mask_conversion_pattern)
17709 (vect_try_gather_scatter_pattern): Likewise.
17711 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17713 * tree-vect-patterns.c (vect_get_internal_def): New function.
17714 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
17715 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
17716 (search_type_for_mask_1): Use it.
17718 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17720 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
17721 redundant WIDEN_SUM_EXPR handling.
17722 (vect_recog_sad_pattern): Likewise.
17724 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17726 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
17727 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
17728 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
17729 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
17730 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
17732 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17734 * tree-vect-stmts.c (vectorizable_call): Make sure that we
17735 use the stmt_vec_info of the original bb statement for the
17736 new zero assignment, even if the call is part of a pattern.
17738 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17740 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
17741 that the sequence is attached to the original statement rather
17742 than the pattern statement.
17743 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
17744 PATTERN_DEF_SEQ from the original statement rather than
17745 the main pattern statement.
17746 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
17747 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
17748 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
17750 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17752 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
17753 definition statements before the early exit for statements that aren't
17755 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
17757 (vect_transform_loop): ...here. Process pattern definition
17758 statements without first checking whether the main pattern
17759 statement is live or relevant.
17761 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
17763 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
17764 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
17766 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
17768 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
17769 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
17770 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
17771 (expand_block_compare): Change select_block_compare_mode call.
17772 (expand_strncmp_align_check): Use new functions, fix comment.
17773 (emit_final_str_compare_gpr): New function.
17774 (expand_strn_compare): Refactor and clean up code.
17775 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
17777 2018-06-19 Tony Reix <tony.reix@atos.com>
17778 Damien Bergamini <damien.bergamini@atos.com>
17779 David Edelsohn <dje.gcc@gmail.com>
17781 * collect2.c (static_obj): New variable.
17782 (static_libs): New variable.
17783 (is_in_list): Uncomment declaration.
17784 (main): Track AIX libraries linked statically.
17785 (is_in_list): Uncomment definition.
17786 (scan_prog_file): Don't add AIX shared libraries initializer
17787 to constructor list if linking statically.
17789 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
17791 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
17793 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
17795 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
17797 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
17800 2018-06-19 Martin Liska <mliska@suse.cz>
17802 * config/i386/i386.c (ix86_can_inline_p): Do not use
17803 ipa_fn_summaries::get_create.
17804 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
17806 (devirtualization_time_bonus): Likewise.
17807 (ipcp_propagate_stage): Likewise.
17808 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
17809 (edge_set_predicate): Likewise.
17810 (evaluate_conditions_for_known_args): Likewise.
17811 (evaluate_properties_for_edge): Likewise.
17812 (ipa_call_summary::reset): Tranform to ...
17813 (ipa_call_summary::~ipa_call_summary): ... this.
17814 (ipa_fn_summary::reset): Transform to ...
17815 (ipa_fn_summary::~ipa_fn_summary): ... this.
17816 (ipa_fn_summary_t::remove): Rename to ...
17817 (ipa_fn_summary_t::remove_callees): ... this.
17818 (ipa_fn_summary_t::duplicate): Use placement new
17819 instead of memory copy.
17820 (ipa_call_summary_t::duplicate): Likewise.
17821 (ipa_call_summary_t::remove): Remove.
17822 (dump_ipa_call_summary): Change get_create to get.
17823 (ipa_dump_fn_summary): Dump only when summary exists.
17824 (analyze_function_body): Use symbol_summary::get instead
17826 (compute_fn_summary): Likewise.
17827 (estimate_edge_devirt_benefit): Likewise.
17828 (estimate_edge_size_and_time): Likewise.
17829 (inline_update_callee_summaries): Likewise.
17830 (remap_edge_change_prob): Likewise.
17831 (remap_edge_summaries): Likewise.
17832 (ipa_merge_fn_summary_after_inlining): Likewise.
17833 (write_ipa_call_summary): Likewise.
17834 (ipa_fn_summary_write): Likewise.
17835 (ipa_free_fn_summary): Likewise.
17836 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
17837 (struct ipa_call_summary): Likewise.
17838 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
17840 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
17841 (estimate_size_after_inlining): Likewise.
17842 (estimate_growth): Likewise.
17843 (growth_likely_positive): Likewise.
17844 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
17845 (inline_call): Likewise.
17846 * ipa-inline.c (caller_growth_limits): Likewise.
17847 (can_inline_edge_p): Likewise.
17848 (can_inline_edge_by_limits_p): Likewise.
17849 (compute_uninlined_call_time): Likewise.
17850 (compute_inlined_call_time): Likewise.
17851 (want_inline_small_function_p): Likewise.
17852 (edge_badness): Likewise.
17853 (update_caller_keys): Likewise.
17854 (update_callee_keys): Likewise.
17855 (inline_small_functions): Likewise.
17856 (inline_to_all_callers_1): Likewise.
17857 (dump_overall_stats): Likewise.
17858 (early_inline_small_functions): Likewise.
17859 (early_inliner): Likewise.
17860 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
17861 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
17862 * ipa-pure-const.c (malloc_candidate_p): Likewise.
17863 * ipa-split.c (execute_split_functions): Likewise.
17864 * symbol-summary.h: Likewise.
17865 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
17867 2018-06-19 Richard Biener <rguenther@suse.de>
17869 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
17870 (vectorize_loops): ... here. Fix dbgcnt handling.
17871 (try_vectorize_loop): Wrap try_vectorize_loop_1.
17873 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
17876 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
17877 ieee128 argument takes up only one (vector) register, not two (floating
17880 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
17882 * gimplify.c (gimplify_init_constructor): Really never clear for an
17883 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
17885 2018-06-19 Richard Biener <rguenther@suse.de>
17887 PR tree-optimization/86179
17888 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
17889 after failed recognition.
17891 2018-06-18 Martin Sebor <msebor@redhat.com>
17893 PR middle-end/85602
17894 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
17895 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
17896 Handle integer subtraction.
17897 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
17898 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
17900 2018-06-18 David Malcolm <dmalcolm@redhat.com>
17902 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
17903 param from rtx to rtx_insn *.
17904 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
17906 (frv_ifcvt_modify_insn): Likwise.
17907 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
17908 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
17909 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
17910 as_a <rtx_insn *> cast to local "unprotected_region" once
17911 it's been established that it's not NULL or pc_rtx.
17912 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
17913 param "sethi" from rtx to rtx_insn *.
17914 (nds32_group_float_insns): Likewise for param "insn".
17915 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
17917 (vax_output_int_subtract): Likewise.
17918 * config/vax/vax.c (vax_output_int_add): Likewise for param
17920 (vax_output_int_subtract): Likewise.
17921 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
17922 (emit_pattern_after): Likewise for param "after".
17923 (emit_insn_after): Likewise.
17924 (emit_jump_insn_after): Likewise.
17925 (emit_call_insn_after): Likewise.
17926 (emit_debug_insn_after): Likewise.
17927 (emit_pattern_before): Likewise for param "before".
17928 (emit_insn_before): Likewise.
17929 (emit_jump_insn_before): Likewise.
17930 * final.c (get_insn_template): Likewise for param "insn", removing
17932 * output.h (get_insn_template): Likewise for 2nd param.
17933 * rtl.h (emit_insn_before): Likewise.
17934 (emit_jump_insn_before): Likewise.
17935 (emit_debug_insn_before_noloc): Likewise.
17936 (emit_insn_after): Likewise.
17937 (emit_jump_insn_after): Likewise.
17938 (emit_call_insn_after): Likewise.
17939 (emit_debug_insn_after): Likewise.
17940 (set_insn_deleted): Likewise for param.
17942 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
17945 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
17946 floating point modes, so that IFmode is numerically greater than
17947 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
17948 to declare the ordering. This prevents IFmode from being
17949 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
17950 machine. Include rs6000-modes.h to share the fractional values
17951 between genmodes* and the rest of the compiler.
17952 (IFmode): Likewise.
17953 (KFmode): Likewise.
17954 (TFmode): Likewise.
17955 * config/rs6000/rs6000-modes.h: New file.
17956 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
17957 meaning of rs6000_long_double_size so that 126..128 selects an
17958 appropriate 128-bit floating point type.
17959 (rs6000_option_override_internal): Likewise.
17960 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
17961 (TARGET_LONG_DOUBLE_128): Change the meaning of
17962 rs6000_long_double_size so that 126..128 selects an appropriate
17963 128-bit floating point type.
17964 (LONG_DOUBLE_TYPE_SIZE): Update comment.
17965 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
17966 source and destination to match the standard usage.
17967 (truncifkf2): Likewise.
17968 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
17969 ISA 2.07 to use an explicit clobber, instead of passing in a
17971 (copysign<mode>3_soft): Likewise.
17973 2018-06-18 David Malcolm <dmalcolm@redhat.com>
17975 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
17976 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
17977 (vect_slp_analyze_instance_dependence): Likewise.
17978 (vect_enhance_data_refs_alignment): Likewise.
17979 (vect_analyze_data_refs_alignment): Likewise.
17980 (vect_slp_analyze_and_verify_instance_alignment
17981 (vect_analyze_data_ref_accesses): Likewise.
17982 (vect_prune_runtime_alias_test_list): Likewise.
17983 (vect_analyze_data_refs): Likewise.
17984 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
17985 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
17986 (vect_analyze_scalar_cycles_1): Likewise.
17987 (vect_get_loop_niters): Likewise.
17988 (vect_analyze_loop_form_1): Likewise.
17989 (vect_update_vf_for_slp): Likewise.
17990 (vect_analyze_loop_operations): Likewise.
17991 (vect_analyze_loop): Likewise.
17992 (vectorizable_induction): Likewise.
17993 (vect_transform_loop): Likewise.
17994 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
17995 * tree-vect-slp.c (vect_analyze_slp): Likewise.
17996 (vect_make_slp_decision): Likewise.
17997 (vect_detect_hybrid_slp): Likewise.
17998 (vect_slp_analyze_operations): Likewise.
17999 (vect_slp_bb): Likewise.
18000 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
18001 (vectorizable_bswap): Likewise.
18002 (vectorizable_call): Likewise.
18003 (vectorizable_simd_clone_call): Likewise.
18004 (vectorizable_conversion): Likewise.
18005 (vectorizable_assignment): Likewise.
18006 (vectorizable_shift): Likewise.
18007 (vectorizable_operation): Likewise.
18008 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
18010 2018-06-18 Martin Sebor <msebor@redhat.com>
18012 PR tree-optimization/81384
18013 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
18014 * builtins.c (expand_builtin_strnlen): New function.
18015 (expand_builtin): Call it.
18016 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
18017 * builtins.def (BUILT_IN_STRNLEN): New.
18018 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
18019 Warn for bounds in excess of maximum object size.
18020 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
18021 single-value ranges. Handle strnlen.
18022 (handle_builtin_strlen): Handle strnlen.
18023 (strlen_check_and_optimize_stmt): Same.
18024 * doc/extend.texi (Other Builtins): Document strnlen.
18026 2018-06-18 Maya Rashish <coypu@sdf.org>
18028 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
18029 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
18030 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
18032 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
18034 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
18036 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18038 * tree.c (escaped_string::escape): Replace cast to char * by
18039 const_cast<char *> (unescaped).
18041 2018-06-18 Nick Clifton <nickc@redhat.com>
18044 * tree.c (escaped_string): New class. Converts an unescaped
18045 string into its escaped equivalent.
18046 (warn_deprecated_use): Use the new class to convert the
18047 deprecation message, if present.
18048 (test_escaped_strings): New self test.
18049 (test_c_tests): Add test_escaped_strings.
18050 * doc/extend.texi (deprecated): Add a note that the
18051 deprecation message is affected by the -fmessage-length
18052 option, and that control characters will be escaped.
18053 (#pragma GCC error): Document this pragma.
18054 (#pragma GCC warning): Likewise.
18055 * doc/invoke.texi (-fmessage-length): Document this option's
18056 effect on the #warning and #error preprocessor directives and
18057 the deprecated attribute.
18059 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
18061 * tree.c (decl_value_expr_lookup): Revert latest change.
18062 (decl_value_expr_insert): Likewise.
18064 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
18066 * gimplify.c (nonlocal_vlas): Delete.
18067 (nonlocal_vla_vars): Likewise.
18068 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
18070 (gimplify_body): Do not create and destroy nonlocal_vlas.
18071 * tree-nested.c: Include diagnostic.h.
18072 (use_pointer_in_frame): Tweak.
18073 (lookup_field_for_decl): Add assertion and declare the transformation.
18074 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
18075 internal error when the reference is in a wrong context. Do not
18076 create a debug decl by default.
18077 (note_nonlocal_block_vlas): Delete.
18078 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
18079 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
18080 create a debug decl by default.
18081 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
18082 call is in a wrong context.
18083 (fixup_vla_decls): New function.
18084 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
18085 debug variables were created.
18086 * tree.c (decl_value_expr_lookup): Add checking assertion.
18087 (decl_value_expr_insert): Likewise.
18089 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
18091 PR middle-end/82479
18092 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
18093 * tree-scalar-evolution.c (interpret_expr): Likewise.
18094 (expression_expensive_p): Likewise.
18095 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
18096 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
18097 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
18098 (ssa_defined_by_minus_one_stmt_p): New.
18100 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
18102 PR middle-end/64946
18103 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
18104 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
18105 * dojump.c (do_jump): Likewise.
18106 * expr.c (expand_expr_real_2): Check operand type's sign.
18107 * fold-const.c (const_unop): Handle ABSU_EXPR.
18108 (fold_abs_const): Likewise.
18109 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
18110 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
18111 (strip_sign_op_1): Likesise.
18112 * match.pd: Add new pattern to generate ABSU_EXPR.
18113 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
18114 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
18115 * tree-eh.c (operation_could_trap_helper_p): Likewise.
18116 * tree-inline.c (estimate_operator_cost): Likewise.
18117 * tree-pretty-print.c (dump_generic_node): Likewise.
18118 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
18119 * tree.def (ABSU_EXPR): New.
18121 2018-06-16 Jakub Jelinek <jakub@redhat.com>
18123 PR middle-end/86095
18124 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
18125 documented as preserved for backward compatibility only.
18126 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
18128 PR rtl-optimization/86108
18129 * bb-reorder.c (create_forwarder_block): Renamed to ...
18130 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
18131 jump from new landing pad to the second part.
18132 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
18135 2018-06-15 Jakub Jelinek <jakub@redhat.com>
18137 PR middle-end/85878
18138 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
18139 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
18140 Only call store_expr for halves if the mode is the same.
18142 PR middle-end/86123
18143 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
18144 Fix up comment formatting.
18146 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
18148 * typed-splay-tree.h (typed_splay_tree::remove): New function.
18149 (typed_splay_tree::closure,
18150 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
18151 (typed_splay_tree::typed_splay_tree,
18152 typed_splay_tree::operator =): Declared private.
18153 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
18154 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
18155 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
18156 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
18157 typed_splay_tree::splay_tree_splay,
18158 typed_splay_tree::splay_tree_foreach_helper,
18159 typed_splay_tree::splay_tree_insert,
18160 typed_splay_tree::splay_tree_remove,
18161 typed_splay_tree::splay_tree_lookup,
18162 typed_splay_tree::splay_tree_predecessor,
18163 typed_splay_tree::splay_tree_successor,
18164 typed_splay_tree::splay_tree_min,
18165 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
18166 (typed_splay_tree::root, typed_splay_tree::comp,
18167 typed_splay_tree::delete_key,
18168 typed_splay_tree::delete_value): New data members.
18169 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
18170 typed_splay_tree::remove.
18172 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
18174 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
18175 -mginv and -mno-ginv to the assembler.
18176 * config/mips/mips.opt (-mcrc): New option.
18177 (-mginv): Likewise.
18178 * doc/invoke.text (-mcrc): Document.
18179 (-mginv): Likewise.
18181 2018-06-15 Nick Clifton <nickc@redhat.com>
18184 * tree.c (escaped_string): New class. Converts an unescaped
18185 string into its escaped equivalent.
18186 (warn_deprecated_use): Use the new class to convert the
18187 deprecation message, if present.
18188 (test_escaped_strings): New self test.
18189 (test_c_tests): Add test_escaped_strings.
18190 * doc/extend.texi (deprecated): Add a note that the
18191 deprecation message is affected by the -fmessage-length
18192 option, and that control characters will be escaped.
18193 (#pragma GCC error): Document this pragma.
18194 (#pragma GCC warning): Likewise.
18195 * doc/invoke.texi (-fmessage-length): Document this option's
18196 effect on the #warning and #error preprocessor directives and
18197 the deprecated attribute.
18199 2018-06-15 Richard Biener <rguenther@suse.de>
18201 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
18202 here, also noting vector size used.
18203 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
18204 size used in MSG_OPTIMIZED_LOCATIONS dump.
18205 (pass_slp_vectorize::execute): Adjust.
18207 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
18210 * config/arc/arc.c (arc_return_address_register): Fix
18213 2018-06-15 Richard Biener <rguenther@suse.de>
18215 PR middle-end/86159
18216 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
18217 leave useless conversion stripping to force_gimple_operand_gsi.
18218 (gimplify_build2): Likewise.
18219 (gimplify_build1): Likewise.
18221 2018-06-15 Richard Biener <rguenther@suse.de>
18223 PR middle-end/86076
18224 * tree-cfg.c (move_stmt_op): unshare invariant addresses
18225 before adjusting their block.
18227 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
18229 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
18230 multilibs for *-*-rtems*.
18231 * config/riscv/t-rtems: New file.
18233 2018-06-14 Jakub Jelinek <jakub@redhat.com>
18235 PR middle-end/86122
18236 * match.pd ((A +- CST1) +- CST2): Punt if last resort
18237 unsigned_type_for returns NULL.
18240 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
18241 subregs of multi-word pseudos unless the float mode has word size.
18243 2018-06-14 Richard Biener <rguenther@suse.de>
18245 PR middle-end/86139
18246 * tree-vect-generic.c (build_word_mode_vector_type): Remove
18247 duplicate and harmful type_hash_canon.
18248 * tree.c (type_hash_canon): Assert we didn't find ourselves.
18250 2018-06-14 Richard Biener <rguenther@suse.de>
18253 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
18256 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
18258 * config/rtems.h (STDINT_LONG32): Define.
18260 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
18261 Prachi Godbole <prachi.godbole@imgtec.com>
18263 * config/mips/mips-cpus.def: Define P6600.
18264 * config/mips/mips-tables.opt: Regenerate.
18265 * config/mips/mips.c (mips_ucbranch_type): New enum.
18266 (mips_rtx_cost_data): Add support for P6600.
18267 (mips_issue_rate): Likewise.
18268 (mips_multipass_dfa_lookahead): Likewise.
18269 (mips_avoid_hazard): Likewise.
18270 (mips_reorg_process_insns): Likewise.
18271 (mips_classify_branch_p6600): New function.
18272 * config/mips/mips.h (TUNE_P6600): New define.
18273 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
18274 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
18275 * config/mips/mips.md: Include p6600.md.
18276 (processor): Add p6600.
18277 * config/mips/p6600.md: New file.
18278 * doc/invoke.texi: Add p6600 to supported architectures.
18280 2018-06-13 Martin Sebor <msebor@redhat.com>
18282 PR tree-optimization/86114
18283 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
18285 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
18287 2018-06-13 Richard Biener <rguenther@suse.de>
18289 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
18290 Properly set vector type of the intermediate stmt.
18291 * tree-vect-stmts.c (vectorizable_operation): The destination
18292 var always has vectype_out type.
18294 2018-06-13 Jeff Law <law@redhat.com>
18296 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
18297 integer 0 for argument to print_rtl_with_bb.
18298 (rl78_reorg): Likewise.
18300 2018-06-13 David Malcolm <dmalcolm@redhat.com>
18302 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
18303 from rtx to rtx_insn *.
18304 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
18306 (add_sched_insns_for_speculation): Likewise for local "target",
18307 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
18308 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
18309 from rtx_insn ** to rtx_code_label **.
18310 (reorg_emit_nops): Likewise.
18311 (c6x_reorg): Likewise for local "call_labels".
18312 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
18314 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
18315 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
18316 the loops over LABEL_REFS.
18317 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
18319 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
18320 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
18321 (split_branches): Strengthen local "olabel" from rtx to
18322 rtx_insn *, adding a safe_as_a cast.
18323 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
18325 (add_insn_after): Likewise for first two params.
18326 (add_insn_before): Likewise.
18327 (remove_insn): Likewise for param.
18328 (emit_pattern_before_noloc): Likewise for second and third params.
18329 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
18330 (emit_call_insn_before_noloc): Likewise.
18331 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
18333 (emit_barrier_before): Likewise.
18334 (emit_label_before): Strengthen "label" param from "rtx" to
18335 "rtx_code_label *". Strengthen "before" param from "rtx" to
18337 (emit_insn_after_1): Strengthen "after" param from "rtx" to
18339 (emit_pattern_after_noloc): Likewise.
18340 (emit_insn_after_noloc): Likewise.
18341 (emit_jump_insn_after_noloc): Likewise.
18342 (emit_call_insn_after_noloc): Likewise.
18343 (emit_debug_insn_after_noloc): Likewise.
18344 (emit_barrier_after): Likewise.
18345 (emit_label_after): Likewise for both params.
18346 (emit_pattern_after_setloc): Likewise for "after" param. Convert
18347 "loc" param from "int" to "location_t".
18348 (emit_insn_after_setloc): Likewise.
18349 (emit_jump_insn_after_setloc): Likewise.
18350 (emit_call_insn_after_setloc): Likewise.
18351 (emit_debug_insn_after_setloc): Likewise.
18352 (emit_pattern_before_setloc): Likewise for "before" param. Convert
18353 "loc" param from "int" to "location_t".
18354 (emit_pattern_before): Convert NULL_RTX to NULL.
18355 (emit_insn_before_setloc): Convert "loc" param from "int" to
18357 (emit_jump_insn_before_setloc): Likewise.
18358 (emit_call_insn_before_setloc): Likewise.
18359 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
18360 rtx_insn *. Convert "loc" param from "int" to "location_t".
18361 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
18362 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
18363 Convert 3rd param from "int" to "location_t".
18364 (emit_barrier_before, emit_barrier_after, next_real_insn):
18365 Strengthen param from rtx to rtx_insn *.
18366 (emit_label_before): Strengthen 1st param from "rtx" to
18367 "rtx_code_label *". Strengthen 2nd param from "rtx" to
18369 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
18370 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
18371 Strengthen 2nd param from "rtx" to "rtx_insn *".
18372 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
18373 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
18374 Likewise. Convert 3rd param from "int" to "location_t".
18375 (emit_label_after): Strengthen 1st param from "rtx" to
18376 "rtx_code_label *".
18377 (next_real_insn, remove_insn): Strengthen param from "rtx" to
18379 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
18380 from "rtx" to "rtx_insn *".
18382 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
18384 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
18385 bodies streamed in with -Q.
18386 * dumpfile.c (dump_files): Add lto-stream-out dump file.
18387 * dumpfile.h (tree_dump_index): Add lto_stream_out.
18388 * gimple-streamer-out.c: Include gimple-pretty-print.h
18389 (output_bb): Dump stmts streamed.
18390 * lto-section-out.c: Include print-tree.h
18391 (lto_begin_section): Dump sections created.
18392 (lto_output_decl_index): Dump decl encoded.
18393 * lto-streamer-out.c: Include print-tree.h
18394 (create_output_block): Dump output block created.
18395 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
18396 (output_function): Dump function output.
18397 (output_constructor): Dump constructor streamed.
18398 (write_global_stream): Output indexes encoded.
18399 (produce_asm_for_decls): Dump streams encoded.
18400 * lto-streamer.c (streamer_dump_file): New global var.
18401 * lto-streamer.h (streamer_dump_file): Declare.
18402 * passes.c (ipa_write_summaries): Initialize streamer dump.
18403 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
18406 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
18409 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
18410 offsets for register save directives. Emit a second batch of save
18411 directives, if need be, when the function accesses prior frames.
18413 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18415 * config/arc/fpu.md (fmasf4): Force operand to register.
18416 (fnmasf4): Likewise.
18418 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18420 * config/arc/arc-protos.h (arc_pad_return): Remove.
18421 * config/arc/arc.c (machine_function): Remove force_short_suffix
18423 (arc_print_operand): Adjust printing of '&'.
18424 (arc_verify_short): Remove conditional printing of short suffix.
18425 (arc_final_prescan_insn): Remove reference to size_reason.
18426 (pad_return): New function.
18427 (arc_reorg): Call pad_return.
18428 (arc_pad_return): Remove.
18429 (arc_init_machine_status): Remove reference to force_short_suffix.
18430 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
18431 (attr length): When attribute iscompact is true force to 2
18432 regardless; in the case of maybe check if we want to force the
18433 instruction to have 4 bytes length.
18434 (nopv): Change it to generate 4 byte long nop as well.
18435 (blockage): New pattern.
18436 (simple_return): Remove call to arc_pad_return.
18437 (p_return_i): Likewise.
18439 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18441 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
18443 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18445 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
18448 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18450 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
18451 for ARC700 and ARCv2.
18453 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
18456 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
18457 operands[2] instead of operands[1].
18460 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
18462 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
18463 case, check whether the outer register overlaps an unallocatable
18464 register, not just whether it fits the required class.
18466 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
18468 * poly-int.h (can_div_trunc_p): Add new overload in which all values
18470 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
18471 (memrefs_conflict_p): Likewise.
18472 (init_alias_analysis): Likewise.
18473 * cfgexpand.c (expand_debug_expr): Likewise.
18474 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
18475 * cse.c (fold_rtx): Likewise.
18476 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
18477 * expr.c (emit_block_move_hints): Likewise.
18478 (clear_storage_hints, push_block, emit_push_insn): Likewise.
18479 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
18480 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
18481 (emit_group_store): Likewise.
18482 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
18483 to read the PRE/POST_MODIFY increment.
18484 * calls.c (store_one_arg): Use strip_offset.
18485 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
18487 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
18489 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
18490 (simplify_binary_operation_1): Extend CONST_INT handling to
18492 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
18493 than a HOST_WIDE_INT.
18494 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
18496 (adjust_mems, add_stores): Update accodingly.
18497 (vt_canonicalize_addr): Track polynomial offsets.
18498 (emit_note_insn_var_location): Likewise.
18499 (vt_add_function_parameter): Likewise.
18500 (vt_initialize): Likewise.
18502 2018-06-12 Jeff Law <law@redhat.com>
18504 * config.gcc (alpha*-*-freebsd*): Remove.
18505 * config/alpha/freebsd.h: Remove.
18507 2018-06-12 David Malcolm <dmalcolm@redhat.com>
18510 * spellcheck-tree.c (levenshtein_distance): Rename to...
18511 (get_edit_distance): ...this, and update for underlying renaming.
18512 * spellcheck-tree.h (levenshtein_distance): Rename to...
18513 (get_edit_distance): ...this.
18514 * spellcheck.c (levenshtein_distance): Rename to...
18515 (get_edit_distance): ...this. Convert from Levenshtein distance
18516 to Damerau-Levenshtein distance by supporting transpositions of
18517 adjacent characters. Rename "v1" to "v_next" and "v0" to
18519 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
18520 (selftest::test_edit_distance_unit_test_oneway): ...this, and
18521 update for underlying renaming.
18522 (selftest::levenshtein_distance_unit_test): Rename to...
18523 (selftest::test_get_edit_distance_unit): ...this, and update for
18524 underlying renaming.
18525 (selftest::test_find_closest_string): Add example from PR 69968
18526 where transposition helps
18527 (selftest::test_metric_conditions): Update for renaming.
18528 (selftest::test_metric_conditions): Likewise.
18529 (selftest::spellcheck_c_tests): Likewise.
18530 * spellcheck.h (levenshtein_distance): Rename both overloads to...
18531 (get_edit_distance): ...this.
18532 (best_match::consider): Update for renaming.
18534 2018-06-12 Martin Sebor <msebor@redhat.com>
18536 PR tree-optimization/85259
18537 * builtins.c (compute_objsize): Handle constant offsets.
18538 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
18539 true iff a warning has been issued.
18540 * gimple.h (gimple_nonartificial_location): New function.
18541 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
18542 gimple_nonartificial_location and handle -Wno-system-headers.
18543 (handle_builtin_stxncpy): Same.
18545 2018-06-12 Martin Sebor <msebor@redhat.com>
18548 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
18550 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18552 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18553 BUILTIN_VEC_XST entries for pointer to double and long long.
18555 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
18558 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
18560 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
18563 2018-06-12 Martin Liska <mliska@suse.cz>
18565 * doc/options.texi: Document IntegerRange.
18567 2018-06-12 Martin Liska <mliska@suse.cz>
18569 * config/i386/i386.opt: Make MPX-related options as Deprecated.
18570 * opt-functions.awk: Handle Deprecated flag.
18571 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
18573 (read_cmdline_option): Report warning for a deprecated option.
18574 * opts.h (struct cl_option): Add new field cl_deprecated.
18575 (CL_ERR_DEPRECATED): New.
18577 2018-06-12 Martin Liska <mliska@suse.cz>
18579 * doc/options.texi: Document Deprecated option flag.
18581 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18583 * config/arc/arc-arch.h (arc_extras): New enum.
18584 (arc_cpu_t):Add field extra.
18585 (arc_cpu_types): Consider the extras.
18586 * config/arc/arc-cpus.def: Add extras info.
18587 * config/arc/arc-opts.h (processor_type): Consider extra field.
18588 * config/arc/arc.c (arc_override_options): Handle extra field.
18590 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18592 * config/arc/arc-arch.h: Update ARC_OPTX macro.
18593 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
18595 * config/arc/arc.c (arc_init): Update pic warning.
18596 (irq_range): Update irq range parsing warnings.
18597 (arc_override_options): Update various warning messages.
18598 (arc_handle_aux_attribute): Likewise.
18600 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
18602 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
18604 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
18606 * doc/sourcebuild.texi: Document usage of line number 0 in verify
18607 compiler messages directives.
18609 2018-06-12 Matthew Fortune <mfortune@gmail.com>
18611 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
18612 * config/mips/mips-tables.opt: Regenerate.
18613 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
18615 * doc/invoke.texi: Document -march=i6500.
18617 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
18619 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
18620 (i6400_gpmul): Add cpu_unit.
18621 (i6400_gpdiv): Likewise.
18622 (i6400_msa_add_d): Update reservations.
18623 (i6400_msa_int_add) Likewise.
18624 (i6400_msa_short_logic3) Likewise.
18625 (i6400_msa_short_logic2) Likewise.
18626 (i6400_msa_short_logic) Likewise.
18627 (i6400_msa_move) Likewise.
18628 (i6400_msa_cmp) Likewise.
18629 (i6400_msa_short_float2) Likewise.
18630 (i6400_msa_div_d) Likewise.
18631 (i6400_msa_long_logic1) Likewise.
18632 (i6400_msa_long_logic2) Likewise.
18633 (i6400_msa_mult) Likewise.
18634 (i6400_msa_long_float2) Likewise.
18635 (i6400_msa_long_float4) Likewise.
18636 (i6400_msa_long_float5) Likewise.
18637 (i6400_msa_long_float8) Likewise.
18638 (i6400_fpu_fadd): Include frint type.
18639 (i6400_fpu_store): New define_insn_reservation.
18640 (i6400_fpu_load): Likewise.
18641 (i6400_fpu_move): Likewise.
18642 (i6400_fpu_fcmp): Likewise.
18643 (i6400_fpu_fmadd): Likewise.
18644 (i6400_int_mult): Include imul3nc type and update reservation.
18645 (i6400_int_div): Include idiv3 type and update reservation.
18646 (i6400_int_load): Update to check type not move_type.
18647 (i6400_int_store): Likewise.
18648 (i6400_int_prefetch): Set zero latency.
18650 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
18652 * gcc.c: Document new %@{...} sequence.
18653 (LINK_COMMAND_SPEC): Use it for the -L switches.
18654 (cpp_unique_options): Use it for the -I switches.
18655 (at_file_argbuf): New global variable.
18656 (in_at_file): Likewise.
18657 (alloc_args): Create at_file_argbuf.
18658 (clear_args): Truncate at_file_argbuf.
18659 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
18660 (open_at_file): New function.
18661 (close_at_file): Likewise.
18662 (create_at_file): Delete.
18663 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
18666 (validate_switches_from_spec): Deal with %@{...} sequence.
18667 (validate_switches): Likewise.
18668 (driver::finalize): Call clear_args.
18670 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
18672 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
18674 2018-06-11 Martin Sebor <msebor@redhat.com>
18676 * doc/invoke.texi (-Wall): List -Wc++17-compat.
18677 (Wno-class-memaccess): Add @opindex.
18678 (Wno-templates, Wno-multiple-inheritance): Same.
18679 (Wno-virtual-inheritance, Wno-namespaces): Same.
18680 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
18681 (Wno-format-overflow, Wno-format-truncation): Same.
18682 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
18683 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
18684 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
18685 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
18686 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
18687 (Wno-misspelled-isr): Same.
18689 2018-06-11 Martin Sebor <msebor@redhat.com>
18691 * PR tree-optimization/86083
18692 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
18694 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
18696 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
18698 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
18701 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
18702 on the correct operand.
18703 (*movdi_internal64): Ditto.
18705 2018-06-11 Martin Liska <mliska@suse.cz>
18707 PR tree-optimization/86089
18708 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
18710 2018-06-11 Julia Koval <julia.koval@intel.com>
18712 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
18713 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
18714 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
18716 2018-06-11 Olivier Hainque <hainque@adacore.com>
18718 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
18719 for Ada with strict dwarf2.
18721 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
18724 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
18727 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
18729 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
18731 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
18733 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
18736 2018-06-08 Martin Liska <mliska@suse.cz>
18738 * tree-cfg.h (debug_function): Fix argument type to match
18741 2018-06-08 Martin Liska <mliska@suse.cz>
18743 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
18744 Remove usage of MPX-related (and removed) fields.
18745 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
18747 2018-06-08 David Malcolm <dmalcolm@redhat.com>
18749 * cfg.c (debug): Use TDF_NONE rather than 0.
18750 * cfghooks.c (debug): Likewise.
18751 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
18752 (struct dump_option_value_info): Convert to...
18753 (struct kv_pair): ...this template type.
18754 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
18756 (optinfo_verbosity_options): Likewise.
18757 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
18759 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
18760 than int for "optgroup_flags" param.
18761 (dump_generic_expr_loc): Use dump_flags_t rather than int for
18763 (dump_dec): Likewise.
18764 (dump_finish): Use TDF_NONE rather than 0.
18765 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
18766 rather than int for "optgroup_flags" param. Use TDF_NONE rather
18767 than 0. Update for change to option_ptr.
18768 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
18769 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
18770 0. Update for changes to optinfo_verbosity_options and
18772 (opt_info_switch_p): Convert optgroup_flags from int to
18774 (dump_basic_block): Use dump_flags_t rather than int
18775 for "dump_kind" param.
18776 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
18777 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
18778 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
18779 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
18780 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
18781 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
18782 TDF_NONE): Convert from macros to...
18783 (enum dump_flag): ...this new enum.
18784 (dump_flags_t): Update to use enum.
18785 (operator|, operator&, operator~, operator|=, operator&=):
18786 Implement for dump_flags_t.
18787 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
18788 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
18789 Convert from macros to...
18790 (enum optgroup_flag): ...this new enum.
18791 (optgroup_flags_t): New typedef.
18792 (operator|, operator|=): Implement for optgroup_flags_t.
18793 (struct dump_file_info): Convert field "alt_flags" to
18794 dump_flags_t. Convert field "optgroup_flags" to
18796 (dump_basic_block): Use dump_flags_t rather than int for param.
18797 (dump_generic_expr_loc): Likewise.
18798 (dump_dec): Likewise.
18799 (dump_register): Convert param "optgroup_flags" to
18801 (opt_info_enable_passes): Likewise.
18802 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
18804 * gimple-pretty-print.c (debug): Likewise.
18805 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
18806 (merged_store_group::apply_stores): Likewise.
18807 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
18808 * gimple.c (verify_gimple_pp): Likewise.
18809 * graphite-poly.c (print_pbb_body): Likewise.
18810 * passes.c (pass_manager::register_one_dump_file): Convert
18811 local "optgroup_flags" to optgroup_flags_t.
18812 * print-tree.c (print_node): Use TDF_NONE rather than 0.
18814 (debug_body): Likewise.
18815 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
18816 to optgroup_flags_t.
18817 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
18819 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
18820 (convert_mult_to_fma): Likewise.
18821 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
18822 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
18823 * tree-vect-data-refs.c (dump_lower_bound): Convert param
18824 "dump_kind" to dump_flags_t.
18826 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
18828 * config/rs6000/rs6000.c (min, max): Delete.
18830 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
18832 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
18833 -mabi=spe and -mabi=no-spe.
18835 2018-06-08 Martin Liska <mliska@suse.cz>
18837 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
18838 where we expect an existing summary.
18840 2018-06-08 Martin Liska <mliska@suse.cz>
18842 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
18843 * ipa-inline.h (estimate_edge_growth): Likewise.
18845 2018-06-08 Martin Liska <mliska@suse.cz>
18847 * cgraph.c (function_version_hasher::hash): Use
18848 cgraph_node::get_uid ().
18849 (function_version_hasher::equal):
18850 * cgraph.h (cgraph_node::get_uid): New method.
18851 * ipa-inline.c (update_caller_keys): Use
18852 cgraph_node::get_uid ().
18853 (update_callee_keys): Likewise.
18854 * ipa-utils.c (searchc): Likewise.
18855 (ipa_reduced_postorder): Likewise.
18856 * lto-cgraph.c (input_node): Likewise.
18857 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
18858 * symbol-summary.h (symtab_insertion): Likewise.
18859 (symtab_removal): Likewise.
18860 (symtab_duplication): Likewise.
18861 * tree-pretty-print.c (dump_function_header): Likewise.
18862 * tree-sra.c (convert_callers_for_node): Likewise.
18864 2018-06-08 Martin Liska <mliska@suse.cz>
18866 * cgraph.c (symbol_table::create_edge): Always assign a new
18868 (symbol_table::free_edge): Do not recycle numbers.
18869 * cgraph.h (cgraph_edge::get): New method.
18870 * symbol-summary.h (symtab_removal): Use it.
18871 (symtab_duplication): Likewise.
18872 (call_summary::hashable_uid): Remove.
18874 2018-06-08 Martin Liska <mliska@suse.cz>
18876 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
18877 (initialize_growth_caches): Remove.
18878 (free_growth_caches): Likewise.
18879 (do_estimate_edge_time): Use edge_growth_cache.
18880 (do_estimate_edge_size): Likewise.
18881 (do_estimate_edge_hints): Likewise.
18882 * ipa-inline.c (reset_edge_caches): Likewise.
18883 (recursive_inlining): Likewise.
18884 (inline_small_functions): Likewise.
18885 * ipa-inline.h (initialize_growth_caches): Remove.
18886 (estimate_edge_size): Likewise.
18887 (estimate_edge_time): Likewise.
18888 (estimate_edge_hints): Likewise.
18889 (reset_edge_growth_cache): Likewise.
18890 * symbol-summary.h (call_summary::remove): New method.
18892 2018-06-08 Martin Liska <mliska@suse.cz>
18894 * ipa-cp.c (class edge_clone_summary): New summary.
18895 (grow_edge_clone_vectors): Remove.
18896 (ipcp_edge_duplication_hook): Remove.
18897 (class edge_clone_summary_t): New call_summary class.
18898 (ipcp_edge_removal_hook): Remove.
18899 (edge_clone_summary_t::duplicate): New function.
18900 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
18901 (create_specialized_node): Likewise.
18902 (ipcp_driver): Initialize edge_clone_summaries and do not
18905 2018-06-08 Martin Liska <mliska@suse.cz>
18907 * symbol-summary.h (get): New function.
18908 (call_summary::m_initialize_when_cloning): New class member.
18910 2018-06-08 Martin Liska <mliska@suse.cz>
18912 * cgraph.c (cgraph_node::remove): Do not recycle uid.
18913 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
18914 (symbol_table::allocate_cgraph_symbol): Do not set uid.
18915 * passes.c (uid_hash_t): Record removed_nodes by their uids.
18916 (remove_cgraph_node_from_order): Use the removed_nodes set.
18917 (do_per_function_toporder): Likwise.
18918 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
18919 instead of summary_uid.
18920 (symtab_removal): Likewise.
18921 (symtab_duplication): Likewise.
18923 2018-06-08 Martin Liska <mliska@suse.cz>
18925 * ipa-cp.c (ipcp_store_bits_results): Use
18926 ipcp_transformation_sum.
18927 (ipcp_store_vr_results): Likewise.
18928 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
18930 (ipcp_transformation_initialize): ... this.
18931 (ipa_set_node_agg_value_chain):
18932 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
18933 (write_ipcp_transformation_info): Likewise.
18934 (read_ipcp_transformation_info): Likewise.
18935 (ipcp_update_bits): Likewise.
18936 (ipcp_update_vr): Likewise.
18937 (ipcp_transform_function): Likewise.
18938 * ipa-prop.h: Rename ipcp_transformation_summary to
18939 ipcp_transformation.
18940 (class ipcp_transformation_t): New function summary.
18941 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
18942 (ipa_get_agg_replacements_for_node): Likewise.
18944 2018-06-08 Martin Liska <mliska@suse.cz>
18946 * ipa-pure-const.c (struct funct_state_d): Do it class instead
18948 (class funct_state_summary_t): New function_summary class.
18949 (has_function_state): Remove.
18950 (get_function_state): Likewise.
18951 (set_function_state): Likewise.
18952 (add_new_function): Likewise.
18953 (funct_state_summary_t::insert): New function.
18954 (duplicate_node_data): Remove.
18955 (remove_node_data): Remove.
18956 (funct_state_summary_t::duplicate): New function.
18957 (register_hooks): Create new funct_state_summaries.
18958 (pure_const_generate_summary): Use it.
18959 (pure_const_write_summary): Likewise.
18960 (pure_const_read_summary): Likewise.
18961 (propagate_pure_const): Likewise.
18962 (propagate_nothrow): Likewise.
18963 (dump_malloc_lattice): Likewise.
18964 (propagate_malloc): Likewise.
18965 (execute): Do not register hooks, just remove summary
18967 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
18970 2018-06-08 Martin Liska <mliska@suse.cz>
18972 * ipa-reference.c (remove_node_data): Remove.
18973 (duplicate_node_data): Likewise.
18974 (class ipa_ref_var_info_summary_t): New class.
18975 (class ipa_ref_opt_summary_t): Likewise.
18976 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
18977 (get_reference_optimization_summary): Use
18978 ipa_ref_opt_sum_summaries.
18979 (set_reference_vars_info): Remove.
18980 (set_reference_optimization_summary): Likewise.
18981 (ipa_init): Create summaries.
18982 (init_function_info): Use function summary.
18983 (ipa_ref_opt_summary_t::duplicate): New function.
18984 (ipa_ref_opt_summary_t::remove): New function.
18985 (get_read_write_all_from_node): Fix GNU coding style.
18986 (propagate): Use function summary.
18987 (write_node_summary_p): Fix GNU coding style.
18988 (stream_out_bitmap): Likewise.
18989 (ipa_reference_read_optimization_summary): Use function summary.
18990 (ipa_reference_c_finalize): Do not release hooks.
18992 2018-06-08 Martin Liska <mliska@suse.cz>
18994 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
18995 (analyze_function_body): Extract multiple calls of get_create.
18996 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
18997 * ipa-inline.c (recursive_inlining): Use ::get method.
18998 * ipa-inline.h (estimate_edge_growth): Likewise.
19000 2018-06-08 Martin Liska <mliska@suse.cz>
19002 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
19004 (hsa_function_summary::hsa_function_summary): Use the new enum
19006 (hsa_gpu_implementation_p): Use hsa_summaries::get.
19007 * hsa-gen.c (hsa_get_host_function): Likewise.
19008 (get_brig_function_name): Likewise.
19009 * ipa-hsa.c (process_hsa_functions): Likewise.
19010 (ipa_hsa_write_summary): Likewise.
19011 * symbol-summary.h (symtab_duplication): Use ::get function/
19012 (get): New function.
19014 2018-06-08 Martin Liska <mliska@suse.cz>
19016 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
19018 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
19019 (hsa_register_kernel): Likewise.
19020 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
19021 * hsa-gen.c (hsa_get_host_function): Likewise.
19022 (get_brig_function_name): Likewise.
19023 (generate_hsa): Likewise.
19024 (pass_gen_hsail::execute): Likewise.
19025 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
19026 (devirtualization_time_bonus): Likewise.
19027 (ipcp_propagate_stage): Likewise.
19028 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
19029 (edge_set_predicate): Likewise.
19030 (evaluate_conditions_for_known_args): Likewise.
19031 (evaluate_properties_for_edge): Likewise.
19032 (ipa_fn_summary::reset): Likewise.
19033 (ipa_fn_summary_t::duplicate): Likewise.
19034 (dump_ipa_call_summary): Likewise.
19035 (ipa_dump_fn_summary): Likewise.
19036 (analyze_function_body): Likewise.
19037 (compute_fn_summary): Likewise.
19038 (estimate_edge_devirt_benefit): Likewise.
19039 (estimate_edge_size_and_time): Likewise.
19040 (estimate_calls_size_and_time): Likewise.
19041 (estimate_node_size_and_time): Likewise.
19042 (inline_update_callee_summaries): Likewise.
19043 (remap_edge_change_prob): Likewise.
19044 (remap_edge_summaries): Likewise.
19045 (ipa_merge_fn_summary_after_inlining): Likewise.
19046 (ipa_update_overall_fn_summary): Likewise.
19047 (read_ipa_call_summary): Likewise.
19048 (inline_read_section): Likewise.
19049 (write_ipa_call_summary): Likewise.
19050 (ipa_fn_summary_write): Likewise.
19051 (ipa_free_fn_summary): Likewise.
19052 * ipa-hsa.c (process_hsa_functions): Likewise.
19053 (ipa_hsa_write_summary): Likewise.
19054 (ipa_hsa_read_section): Likewise.
19055 * ipa-icf.c (sem_function::merge): Likewise.
19056 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
19057 (do_estimate_edge_time): Likewise.
19058 (estimate_size_after_inlining): Likewise.
19059 (estimate_growth): Likewise.
19060 (growth_likely_positive): Likewise.
19061 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
19062 (inline_call): Likewise.
19063 * ipa-inline.c (caller_growth_limits): Likewise.
19064 (can_inline_edge_p): Likewise.
19065 (can_inline_edge_by_limits_p): Likewise.
19066 (compute_uninlined_call_time): Likewise.
19067 (compute_inlined_call_time): Likewise.
19068 (want_inline_small_function_p): Likewise.
19069 (edge_badness): Likewise.
19070 (update_caller_keys): Likewise.
19071 (update_callee_keys): Likewise.
19072 (recursive_inlining): Likewise.
19073 (inline_small_functions): Likewise.
19074 (inline_to_all_callers_1): Likewise.
19075 (dump_overall_stats): Likewise.
19076 (early_inline_small_functions): Likewise.
19077 (early_inliner): Likewise.
19078 * ipa-inline.h (estimate_edge_growth): Likewise.
19079 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
19080 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
19081 * ipa-prop.h (IPA_NODE_REF): Likewise.
19082 (IPA_EDGE_REF): Likewise.
19083 * ipa-pure-const.c (malloc_candidate_p): Likewise.
19084 (propagate_malloc): Likewise.
19085 * ipa-split.c (execute_split_functions): Likewise.
19086 * symbol-summary.h: Rename get to get_create.
19088 (get_create): Likewise.
19089 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
19091 2018-06-08 Martin Liska <mliska@suse.cz>
19093 * symbol-summary.h (release): Move definition out of class
19095 (symtab_removal): Likewise.
19096 (symtab_duplication): Likewise.
19098 2018-06-08 Martin Liska <mliska@suse.cz>
19100 * symbol-summary.h (function_summary): Move constructor
19101 implementation out of class declaration.
19102 (release): Likewise.
19103 (symtab_insertion): Likewise.
19104 (symtab_removal): Likewise.
19105 (symtab_duplication): Likewise.
19108 2018-06-08 Martin Liska <mliska@suse.cz>
19110 * Makefile.in: Remove support for MPX (macros, related functions,
19111 fields in cgraph_node, ...).
19112 * builtin-types.def (BT_BND): Likewise.
19113 (BT_FN_BND_CONST_PTR): Likewise.
19114 (BT_FN_CONST_PTR_BND): Likewise.
19115 (BT_FN_VOID_PTR_BND): Likewise.
19116 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
19117 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
19118 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
19119 (expand_builtin_mempcpy_with_bounds): Likewise.
19120 (expand_builtin_memset_with_bounds): Likewise.
19121 (expand_builtin_memset_args): Likewise.
19122 (std_expand_builtin_va_start): Likewise.
19123 (expand_builtin): Likewise.
19124 (expand_builtin_with_bounds): Likewise.
19125 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
19126 (DEF_LIB_BUILTIN_CHKP): Likewise.
19127 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
19128 (DEF_CHKP_BUILTIN): Likewise.
19129 (BUILT_IN_MEMCPY): Likewise.
19130 (BUILT_IN_MEMMOVE): Likewise.
19131 (BUILT_IN_MEMPCPY): Likewise.
19132 (BUILT_IN_MEMSET): Likewise.
19133 (BUILT_IN_STPCPY): Likewise.
19134 (BUILT_IN_STRCAT): Likewise.
19135 (BUILT_IN_STRCHR): Likewise.
19136 (BUILT_IN_STRCPY): Likewise.
19137 (BUILT_IN_STRLEN): Likewise.
19138 (BUILT_IN_MEMCPY_CHK): Likewise.
19139 (BUILT_IN_MEMMOVE_CHK): Likewise.
19140 (BUILT_IN_MEMPCPY_CHK): Likewise.
19141 (BUILT_IN_MEMSET_CHK): Likewise.
19142 (BUILT_IN_STPCPY_CHK): Likewise.
19143 (BUILT_IN_STRCAT_CHK): Likewise.
19144 (BUILT_IN_STRCPY_CHK): Likewise.
19145 * calls.c (store_bounds): Likewise.
19146 (emit_call_1): Likewise.
19147 (special_function_p): Likewise.
19148 (maybe_warn_nonstring_arg): Likewise.
19149 (initialize_argument_information): Likewise.
19150 (finalize_must_preallocate): Likewise.
19151 (compute_argument_addresses): Likewise.
19152 (expand_call): Likewise.
19153 * cfgexpand.c (expand_call_stmt): Likewise.
19154 (expand_return): Likewise.
19155 (expand_gimple_stmt_1): Likewise.
19156 (pass_expand::execute): Likewise.
19157 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19158 (cgraph_node::remove): Likewise.
19159 (cgraph_node::dump): Likewise.
19160 (cgraph_node::verify_node): Likewise.
19161 * cgraph.h (chkp_function_instrumented_p): Likewise.
19162 (symtab_node::get_alias_target): Likewise.
19163 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
19164 (cgraph_local_p): Likewise.
19165 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
19166 (cgraph_edge::rebuild_references): Likewise.
19167 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
19168 (walk_polymorphic_call_targets): Likewise.
19169 (cgraph_node::expand_thunk): Likewise.
19170 (symbol_table::output_weakrefs): Likewise.
19171 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
19172 (ix86_handle_option): Likewise.
19173 * config/i386/constraints.md: Likewise.
19174 * config/i386/i386-builtin-types.def (BND): Likewise.
19178 * config/i386/i386-builtin.def (BDESC_END): Likewise.
19179 (BDESC_FIRST): Likewise.
19181 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
19182 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
19183 * config/i386/i386.c (enum reg_class): Likewise.
19184 (ix86_target_string): Likewise.
19185 (ix86_option_override_internal): Likewise.
19186 (ix86_conditional_register_usage): Likewise.
19187 (ix86_valid_target_attribute_inner_p): Likewise.
19188 (ix86_set_indirect_branch_type): Likewise.
19189 (ix86_set_current_function): Likewise.
19190 (ix86_function_arg_regno_p): Likewise.
19191 (init_cumulative_args): Likewise.
19192 (ix86_function_arg_advance): Likewise.
19193 (ix86_function_arg): Likewise.
19194 (ix86_pass_by_reference): Likewise.
19195 (ix86_function_value_regno_p): Likewise.
19196 (ix86_function_value_1): Likewise.
19197 (ix86_function_value_bounds): Likewise.
19198 (ix86_return_in_memory): Likewise.
19199 (ix86_setup_incoming_vararg_bounds): Likewise.
19200 (ix86_va_start): Likewise.
19201 (indirect_thunk_need_prefix): Likewise.
19202 (print_reg): Likewise.
19203 (ix86_print_operand): Likewise.
19204 (ix86_expand_call): Likewise.
19205 (ix86_output_function_return): Likewise.
19206 (reg_encoded_number): Likewise.
19207 (BDESC_VERIFYS): Likewise.
19208 (ix86_init_mpx_builtins): Likewise.
19209 (ix86_init_builtins): Likewise.
19210 (ix86_emit_cmove): Likewise.
19211 (ix86_emit_move_max): Likewise.
19212 (ix86_expand_builtin): Likewise.
19213 (ix86_builtin_mpx_function): Likewise.
19214 (ix86_get_arg_address_for_bt): Likewise.
19215 (ix86_load_bounds): Likewise.
19216 (ix86_store_bounds): Likewise.
19217 (ix86_load_returned_bounds): Likewise.
19218 (ix86_store_returned_bounds): Likewise.
19219 (ix86_class_likely_spilled_p): Likewise.
19220 (ix86_hard_regno_mode_ok): Likewise.
19221 (x86_order_regs_for_local_alloc): Likewise.
19222 (ix86_mitigate_rop): Likewise.
19223 (ix86_bnd_prefixed_insn_p): Likewise.
19224 (ix86_mpx_bound_mode): Likewise.
19225 (ix86_make_bounds_constant): Likewise.
19226 (ix86_initialize_bounds): Likewise.
19227 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
19228 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
19229 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
19230 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
19231 (TARGET_CHKP_BOUND_MODE): Likewise.
19232 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
19233 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
19234 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
19235 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
19236 * config/i386/i386.h (TARGET_MPX): Likewise.
19237 (TARGET_MPX_P): Likewise.
19238 (VALID_BND_REG_MODE): Likewise.
19239 (FIRST_BND_REG): Likewise.
19240 (LAST_BND_REG): Likewise.
19241 (enum reg_class): Likewise.
19242 (BND_REG_P): Likewise.
19243 (BND_REGNO_P): Likewise.
19244 (BNDmode): Likewise.
19245 (ADJUST_INSN_LENGTH): Likewise.
19246 * config/i386/i386.md: Likewise.
19247 * config/i386/i386.opt: Likewise.
19248 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
19249 (defined): Likewise.
19250 (LINK_MPX): Likewise.
19251 (MPX_SPEC): Likewise.
19252 (LIBMPX_SPEC): Likewise.
19253 (LIBMPXWRAPPERS_SPEC): Likewise.
19254 (CHKP_SPEC): Likewise.
19255 * config/i386/predicates.md: Likewise.
19256 * dbxout.c (dbxout_type): Likewise.
19257 * doc/extend.texi: Likewise.
19258 * doc/invoke.texi: Likewise.
19259 * doc/md.texi: Likewise.
19260 * doc/tm.texi: Likewise.
19261 * doc/tm.texi.in: Likewise.
19262 * dwarf2out.c (is_base_type): Likewise.
19263 (gen_formal_types_die): Likewise.
19264 (gen_subprogram_die): Likewise.
19265 (gen_type_die_with_usage): Likewise.
19266 (gen_decl_die): Likewise.
19267 (dwarf2out_late_global_decl): Likewise.
19268 * expr.c (expand_assignment): Likewise.
19269 (emit_storent_insn): Likewise.
19270 (store_expr_with_bounds): Likewise.
19271 (store_expr): Likewise.
19272 (expand_expr_real_1): Likewise.
19273 * expr.h (store_expr_with_bounds): Likewise.
19274 * function.c (use_register_for_decl): Likewise.
19275 (struct bounds_parm_data): Likewise.
19276 (assign_parms_augmented_arg_list): Likewise.
19277 (assign_parm_find_entry_rtl): Likewise.
19278 (assign_parm_is_stack_parm): Likewise.
19279 (assign_parm_load_bounds): Likewise.
19280 (assign_bounds): Likewise.
19281 (assign_parms): Likewise.
19282 (expand_function_start): Likewise.
19283 * gcc.c (CHKP_SPEC): Likewise.
19284 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
19285 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
19286 (wrestrict_dom_walker::check_call): Likewise.
19287 * gimple.c (gimple_build_call_from_tree): Likewise.
19288 * gimple.h (enum gf_mask): Likewise.
19289 (gimple_call_with_bounds_p): Likewise.
19290 (gimple_call_set_with_bounds): Likewise.
19291 * gimplify.c (gimplify_init_constructor): Likewise.
19292 * ipa-cp.c (initialize_node_lattices): Likewise.
19293 (propagate_constants_across_call): Likewise.
19294 (find_more_scalar_values_for_callers_subset): Likewise.
19295 * ipa-hsa.c (process_hsa_functions): Likewise.
19296 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
19297 * ipa-icf.c (sem_function::merge): Likewise.
19298 * ipa-inline.c (early_inliner): Likewise.
19299 * ipa-pure-const.c (warn_function_noreturn): Likewise.
19300 (warn_function_cold): Likewise.
19301 (propagate_pure_const): Likewise.
19302 * ipa-ref.h (enum GTY): Likewise.
19303 * ipa-split.c (find_retbnd): Likewise.
19304 (consider_split): Likewise.
19305 (split_function): Likewise.
19306 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
19307 * ipa.c (walk_polymorphic_call_targets): Likewise.
19308 (symbol_table::remove_unreachable_nodes): Likewise.
19309 (process_references): Likewise.
19310 (cgraph_build_static_cdtor_1): Likewise.
19311 * lto-cgraph.c (lto_output_node): Likewise.
19312 (output_refs): Likewise.
19313 (compute_ltrans_boundary): Likewise.
19314 (input_overwrite_node): Likewise.
19315 (input_node): Likewise.
19316 (input_cgraph_1): Likewise.
19317 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
19318 * passes.c (pass_manager::execute_early_local_passes): Likewise.
19319 (class pass_chkp_instrumentation_passes): Likewise.
19320 (make_pass_chkp_instrumentation_passes): Likewise.
19321 * passes.def: Likewise.
19322 * rtl.h (struct GTY): Likewise.
19323 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
19324 * stor-layout.c (layout_type): Likewise.
19325 * symtab.c: Likewise.
19326 * target.def: Likewise.
19327 * targhooks.c (default_chkp_bound_type): Likewise.
19328 (default_chkp_bound_mode): Likewise.
19329 (default_builtin_chkp_function): Likewise.
19330 (default_chkp_function_value_bounds): Likewise.
19331 (default_chkp_make_bounds_constant): Likewise.
19332 (default_chkp_initialize_bounds): Likewise.
19333 * targhooks.h (default_chkp_bound_type): Likewise.
19334 (default_chkp_bound_mode): Likewise.
19335 (default_builtin_chkp_function): Likewise.
19336 (default_chkp_function_value_bounds): Likewise.
19337 (default_chkp_make_bounds_constant): Likewise.
19338 (default_chkp_initialize_bounds): Likewise.
19339 * toplev.c (compile_file): Likewise.
19340 (process_options): Likewise.
19341 * tree-core.h (DEF_BUILTIN): Likewise.
19342 (DEF_BUILTIN_CHKP): Likewise.
19343 * tree-inline.c (declare_return_variable): Likewise.
19344 (remap_gimple_stmt): Likewise.
19345 (copy_bb): Likewise.
19346 (initialize_inlined_parameters): Likewise.
19347 (expand_call_inline): Likewise.
19348 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
19349 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
19350 (make_pass_ipa_chkp_produce_thunks): Likewise.
19351 (make_pass_chkp): Likewise.
19352 (make_pass_chkp_opt): Likewise.
19353 (make_pass_chkp_instrumentation_passes): Likewise.
19354 * tree-pretty-print.c (dump_generic_node): Likewise.
19355 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
19356 * tree-ssa-dce.c (propagate_necessity): Likewise.
19357 (eliminate_unnecessary_stmts): Likewise.
19358 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
19359 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
19360 * tree-ssa-sccvn.h: Likewise.
19361 * tree-ssa-strlen.c (get_string_length): Likewise.
19362 (valid_builtin_call): Likewise.
19363 (adjust_last_stmt): Likewise.
19364 (handle_builtin_strchr): Likewise.
19365 (handle_builtin_strcpy): Likewise.
19366 (handle_builtin_stxncpy): Likewise.
19367 (handle_builtin_memcpy): Likewise.
19368 (handle_builtin_strcat): Likewise.
19369 (strlen_check_and_optimize_stmt): Likewise.
19370 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
19371 * tree-streamer-in.c: Likewise.
19372 * tree-streamer.c (record_common_node): Likewise.
19373 * tree.c (tree_code_size): Likewise.
19374 (wide_int_to_tree_1): Likewise.
19375 (type_contains_placeholder_1): Likewise.
19376 (build_common_tree_nodes): Likewise.
19377 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
19378 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
19379 (POINTER_BOUNDS_P): Likewise.
19380 (BOUNDED_TYPE_P): Likewise.
19381 (BOUNDED_P): Likewise.
19382 (CALL_WITH_BOUNDS_P): Likewise.
19383 (pointer_bounds_type_node): Likewise.
19384 * value-prof.c (gimple_ic): Likewise.
19385 * var-tracking.c (vt_add_function_parameters): Likewise.
19386 * varasm.c (make_decl_rtl): Likewise.
19387 (assemble_start_function): Likewise.
19388 (output_constant): Likewise.
19389 (maybe_assemble_visibility): Likewise.
19390 * varpool.c (ctor_for_folding): Likewise.
19391 * chkp-builtins.def: Remove.
19392 * ipa-chkp.c: Remove.
19393 * ipa-chkp.h: Remove.
19394 * rtl-chkp.c: Remove.
19395 * rtl-chkp.h: Remove.
19396 * tree-chkp-opt.c: Remove.
19397 * tree-chkp.c: Remove.
19398 * tree-chkp.h: Remove.
19400 2018-06-07 Carl Love <cel@us.ibm.com>
19402 * config/rs6000/vsx.md (vextract_fp_from_shorth,
19403 vextract_fp_from_shortl): Add BE support.
19405 2018-06-07 Paul Koning <ni1d@arrl.net>
19407 * compare-elim.c (try_merge_compare): Don't merge compare if
19408 address contains a side effect.
19409 (try_eliminate_compare): Likewise.
19411 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
19413 * config.gcc: Support "tremont".
19414 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
19415 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19417 * config/i386/i386.c (m_TREMONT): Define.
19418 (processor_target_table): Add "tremont".
19419 (PTA_TREMONT): Define.
19420 (ix86_lea_outperforms): Add TARGET_TREMONT.
19421 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
19422 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
19423 and M_INTEL_GOLDMONT_PLUS.
19424 (fold_builtin_cpu): Add "tremont".
19425 (ix86_add_stmt_cost): Add TARGET_TREMONT.
19426 (ix86_option_override_internal): Add "tremont".
19427 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
19428 (processor_type): Add PROCESSOR_TREMONT.
19429 * config/i386/x86-tune.def: Add m_TREMONT.
19430 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
19432 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
19434 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
19435 symbol defined for msp430i* devices to be lower case.
19437 2018-06-07 Richard Biener <rguenther@suse.de>
19439 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
19440 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
19441 Properly wrap signed arithmetic if overflow wraps.
19443 2018-06-07 Jakub Jelinek <jakub@redhat.com>
19445 PR tree-optimization/69615
19446 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
19447 of a cast from a same precision integral SSA_NAME in a bb dominated
19448 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
19449 cast to utype if rhs2 has already a compatible type.
19451 2018-06-07 Richard Biener <rguenther@suse.de>
19453 PR tree-optimization/85935
19454 * graphite-scop-detection.c (find_params_in_bb): Analyze
19455 condition operands with respect to the correct loop. Assert
19456 the analysis doesn't fail.
19458 2018-06-04 Carl Love <cel@us.ibm.com>
19460 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
19461 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
19462 as it is slightly cheaper.
19463 (first_match_or_eos_index_<mode>):
19464 Calculate index using natural element order.
19465 (first_match_index_<mode>):
19466 Calculate index using natural element order.
19467 (first_match_or_eos_index_<mode>):
19468 Calculate index using natural order.
19469 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
19470 for BE and LE modes.
19471 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
19472 P9V_BUILTIN_VCLZLSBB_V16QI.
19473 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
19476 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
19478 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
19479 indentation and line wrap for many prototypes. Add missing
19480 @smallexample directives around block of prototypes for vec_xl and
19483 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
19485 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
19486 track if we pass or return IEEE 128-bit floating point.
19487 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
19488 C++ mangling that is compatible with GCC 8.1.
19489 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
19490 (init_cumulative_args): Note if we pass or return IEEE 128-bit
19491 floating point types.
19492 (rs6000_function_arg_advance_1): Likewise.
19493 (rs6000_mangle_type): Optionally generate mangled names that match
19494 what GCC 8.1 generated for IEEE 128-bit floating point types.
19495 (rs6000_globalize_decl_name): If we have an external function that
19496 passes or returns IEEE 128-bit types, generate a weak reference
19497 from the mangled name used in GCC 8.1 to the current mangled
19499 (rs6000_init_builtins): Make __ibm128 use the long double type if
19500 long double is IBM extended double. Make __float128 use the long
19501 double type if long double is IEEE 128-bit.
19504 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
19505 macro for __ibm128 built-in functions.
19506 (PACK_IF): Add __ibm128 pack/unpack functions.
19507 (UNPACK_IF): Likewise.
19508 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
19509 enable long double built-in functions if long double is IEEE
19510 128-bit floating point.
19511 (rs6000_invalid_builtin): Update long double built-in function
19513 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
19514 functions, adjust the built-in function to use the long double
19515 built-in function if __ibm128 and long double are the same type.
19516 * doc/extend.texi (PowerPC builtins): Update documention for
19517 __builtin_{,un}pack_longdouble. Add documentation for
19518 __builtin_{,un}pack_ibm128.
19520 2018-06-06 Jim Wilson <jimw@sifive.com>
19522 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
19523 (struct machine_function): New field interrupt_mode.
19524 (riscv_handle_type_attribute): New function. Add forward declaration.
19525 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
19526 (riscv_expand_epilogue): Check interrupt_mode field.
19527 (riscv_set_current_function): Check interrupt attribute args and
19528 set interrupt_mode field.
19529 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
19530 (riscv_sret, riscv_uret): New.
19531 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
19532 new arguments to interrupt attribute.
19534 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
19537 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
19538 Don't handle -mcpu=power8 if -mpower9-vector is also used.
19540 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
19542 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
19543 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
19544 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
19545 several redundant entries.
19547 2018-06-06 David Malcolm <dmalcolm@redhat.com>
19549 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
19550 type from "rtx" to "rtx_insn *".
19551 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
19552 for local "call_insn", removing cast.
19553 (ix86_expand_call): Likewise, introducing a "call_insn" local.
19555 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
19557 PR tree-optimization/86066
19558 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
19559 for BIT_INSERT_EXPR stores.
19561 2018-06-06 Richard Biener <rguenther@suse.de>
19563 PR tree-optimization/86062
19564 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
19565 component refs ontop
19566 of to be offsetted base.
19568 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
19570 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
19571 to be static and remove check on interrupt attribute name.
19573 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
19575 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
19576 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
19578 2018-06-05 Steve Ellcey <sellcey@cavium.com>
19581 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
19583 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
19584 Remove second argument, change how error is called.
19585 (aarch64_layout_arg): Remove second argument from
19586 aarch64_err_no_fpadvsimd call.
19587 (aarch64_init_cumulative_args): Ditto.
19588 (aarch64_gimplify_va_arg_expr): Ditto.
19589 * config/aarch64/aarch64.md (mov<mode>): Ditto.
19591 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
19593 * config/i386/i386.md (simple_return_indirect_internal): New expander.
19594 (*simple_return_indirect_internal<mode>): Rename from
19595 simple_return_indirect_internal. Use W mode iterator.
19596 (rstorssp): New expander.
19597 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
19598 (clrssbsy): New expander.
19599 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
19601 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
19603 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
19605 (cmse_check_pointed_object): Likewise.
19607 2018-06-05 Martin Liska <mliska@suse.cz>
19609 PR gcov-profile/47618
19610 * doc/invoke.texi: Document how -fprofile-dir format
19613 2018-06-05 Richard Biener <rguenther@suse.de>
19615 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
19616 removal pretend DOM info isn't available so we do not update
19617 it and only remove edges, not dominated blocks. Actually free
19618 DOM info in case we removed something. Remove unreachable blocks.
19619 (mfb_keep_latches): Work with either DOM info or marked backedges.
19620 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
19621 first. Mark backedges if DOM info isn't available.
19622 (Re-)compute DOM info after cleanup_control_flow_pre.
19624 2018-06-05 Richard Biener <rguenther@suse.de>
19626 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
19627 (locus_discrim_hasher::hash): Adjust.
19628 (locus_discrim_hasher::equal): Likewise.
19629 (next_discriminator_for_locus): Work on line directly.
19630 (same_line_p): Pass in expanded locus1 as well.
19631 (assign_discriminators): Avoid redundant location expansions.
19633 2018-06-05 Richard Biener <rguenther@suse.de>
19635 PR tree-optimization/86046
19636 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
19637 if required after clearing TREE_ADDRESSABLE.
19639 2018-06-05 Richard Biener <rguenther@suse.de>
19641 PR tree-optimization/86047
19642 * tree-ssa-loop.c (for_each_index): Glob handling of all
19643 decls and constants and really handle all of them.
19645 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19648 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
19649 qualifier_void_pointer and qualifier_const_void_pointer.
19650 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
19651 (arm_init_builtins): Handle the above.
19652 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
19653 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
19654 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
19657 2018-06-05 Martin Liska <mliska@suse.cz>
19659 * auto-profile.c (read_autofdo_file): Do not use
19660 gcov_ctr_summary struct.
19661 (afdo_callsite_hot_enough_for_early_inline): Likewise.
19662 * coverage.c (struct counts_entry): Likewise.
19663 (read_counts_file): Read just single summary entry.
19664 (get_coverage_counts): Use gcov_summary struct.
19665 * coverage.h (get_coverage_counts): Likewise.
19666 * gcov-dump.c (dump_working_sets): Likewise.
19667 (tag_summary): Dump just single summary.
19668 * gcov-io.c (gcov_write_summary): Write just histogram
19670 (gcov_read_summary): Read just single summary.
19671 (compute_working_sets): Use gcov_summary struct.
19672 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
19673 of GCOV_COUNTERS_SUMMABLE.
19674 (GCOV_COUNTERS_SUMMABLE): Remove.
19675 (GCOV_FIRST_VALUE_COUNTER): Replace with
19676 GCOV_COUNTER_V_INTERVAL.
19677 (struct gcov_ctr_summary): Remove.
19678 (struct gcov_summary): Directly use fields of former
19680 (compute_working_sets): Use gcov_summary struct.
19681 * gcov.c (read_count_file): Do not use ctrs fields.
19682 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
19684 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
19686 * profile.c: Likewise.
19687 * profile.h: Likewise.
19689 2018-06-05 Martin Liska <mliska@suse.cz>
19691 PR gcov-profile/84846
19692 * gcov.c (output_lines): Print working directory only
19693 in intermediate format.
19695 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
19697 * config/s390/s390-builtin-types.def: Add void function type.
19698 * config/s390/s390-builtins.def: Use the function type for the
19701 2018-06-04 Jim Wilson <jimw@sifive.com>
19703 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
19705 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
19706 and maybe_eh_return. Change regno to unsigned int. Use new args to
19707 handle EH_RETURN_DATA_REGNO registers properly.
19708 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
19709 (riscv_expand_epilogue): Update comment. Change argument name and
19710 type. Update code to use new name and type. Pass new args to
19711 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
19713 * config/riscv/riscv.md (NORMAL_RETURN): New.
19714 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
19715 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
19716 (eh_return): Call gen_eh_return_internal and emit barrier.
19717 (eh_return_internal): Call riscv_expand_epilogue.
19719 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
19721 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
19722 bit_insertion field and declare can_be_merged_into method.
19723 (merged_store_group::can_be_merged_into): New method.
19724 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
19725 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
19726 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
19728 2018-06-04 Richard Biener <rguenther@suse.de>
19730 PR tree-optimization/85955
19731 * builtins.c (fold_builtin_sincos): Convert pointers to
19732 destination to appropriate type before dereferencing.
19734 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
19736 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
19738 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
19740 * expr.c (expand_expr_real_1): Force the operand into memory if
19741 its TYPE_MODE is BLKmode and if there is no integer mode for
19742 the number of bits being extracted.
19744 2018-06-04 Jakub Jelinek <jakub@redhat.com>
19748 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
19749 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
19751 2018-06-04 Richard Biener <rguenther@suse.de>
19753 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
19754 (cleanup_tree_cfg_noloop): ... single caller. Do
19755 start_recording_case_labels later.
19757 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
19759 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
19760 to _IMMINTRIN_H_INCLUDED.
19761 * config/i386/pconfigintrin.h: Ditto.
19762 * config/i386/waitpkgintrin.h: Ditto.
19763 * config/i386/immintrin.h: Add includes for sgxintrin.h,
19764 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
19765 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
19766 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
19767 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
19768 waitpkgintrin.h and cldemoteintrin.h.
19770 2018-06-04 Richard Biener <rguenther@suse.de>
19772 PR tree-optimization/86038
19773 * tracer.c (find_best_successor): Check probability for
19774 being initialized, bail out if not.
19776 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
19779 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
19780 of bits to ignore when comparing architectures.
19782 2018-06-04 Jakub Jelinek <jakub@redhat.com>
19784 PR tree-optimization/69615
19785 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
19786 maximum or minimum of the type, try to merge it also as if
19787 range1 is + [-, x - 1] or + [x + 1, -].
19790 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
19792 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
19794 PR tree-optimization/86034
19795 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
19796 the unsigned bitfield type in a bit insertion sequence if it does not
19797 have a larger precision than the bitfield size.
19798 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
19800 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
19802 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
19804 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
19806 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
19807 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
19808 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
19809 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
19811 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
19813 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
19814 Disable -fdelete-null-pointer-checks for ELF toolchain.
19816 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
19817 Kito Cheng <kito.cheng@gmail.com>
19819 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
19820 (nds32le-*-*, nds32be-*-*): Integrate checking process.
19821 (nds32*-*-*): Add glibc and uclibc conditions.
19822 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
19823 (TARGET_EXCEPT_UNWIND_INFO): Define.
19824 * config/nds32/elf.h: New file.
19825 * config/nds32/linux.h: New file.
19826 * config/nds32/nds32-elf.opt: New file.
19827 * config/nds32/nds32-linux.opt: New file.
19828 * config/nds32/nds32-fp-as-gp.c
19829 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
19830 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
19832 (nds32_asm_file_end): Ditto.
19833 (nds32_print_operand): Ditto.
19834 (nds32_insert_attributes): Ditto.
19835 (nds32_init_libfuncs): New function.
19836 (TARGET_HAVE_TLS): Define.
19837 (TARGET_INIT_LIBFUNCS): Define.
19838 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
19840 (TARGET_ELF): Apply different mcmodel setting.
19841 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
19842 been migrated into elf.h and linux.h files.
19843 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
19844 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
19845 (mcmodel): The content has been migrated into nds32-elf.opt and
19846 nds32-linux.opt files.
19847 * config/nds32/t-elf: New file.
19848 * config/nds32/t-linux: New file.
19850 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
19851 Shiva Chen <shiva0217@gmail.com>
19853 * config/nds32/constants.md (unspec_volatile_element): Add
19854 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
19855 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
19857 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
19858 (make_pass_nds32_fp_as_gp): Declare.
19859 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
19861 (nds32_asm_function_end_prologue): Remove unused asm output.
19862 (nds32_asm_function_begin_epilogue): Remove unused asm output.
19863 (nds32_asm_file_start): Output necessary fp_as_gp information.
19864 (nds32_option_override): Adjust register usage.
19865 (nds32_expand_prologue): Consider fp_as_gp situation.
19866 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
19867 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
19870 (simple_return): Ditto.
19871 (omit_fp_begin): Output special directive for fp_as_gp.
19872 (omit_fp_end): Output special directive for fp_as_gp.
19873 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
19874 mforbid-fp-as-gp): New options.
19876 2018-06-01 Mark Wielaard <mark@klomp.org>
19878 * dwarf2out.c (dwarf2out_finish): Remove generation of
19879 DW_AT_loclists_base.
19881 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
19883 * gimple-ssa-store-merging.c: Include gimple-fold.h.
19884 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
19885 (struct merged_store_group): Add bit_insertion field.
19886 (dump_char_array): Use standard hexadecimal format.
19887 (merged_store_group::merged_store_group): Set bit_insertion to false.
19888 (merged_store_group::apply_stores): Use optimal buffer size. Deal
19889 with BIT_INSERT_EXPR stores. Move up code updating the mask and
19890 also print the mask in the dump file.
19891 (pass_store_merging::gate): Minor tweak.
19892 (imm_store_chain_info::coalesce_immediate): Fix wrong association
19893 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
19894 stores with INTEGER_CST stores.
19895 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
19896 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
19897 and use it throughout. Generate bit insertion sequences if need be.
19898 (pass_store_merging::process_store): Remove redundant condition.
19899 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
19901 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
19903 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
19904 the 128-bit floating point types. Fix function comment.
19906 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19908 * config/aarch64/aarch64-simd.md
19909 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
19911 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
19914 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
19916 PR tree-optimization/85989
19917 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
19919 (backprop::intersect_uses): Check it when deciding whether this
19920 is a backedge reference.
19921 (backprop::process_block): Add each phi to m_visited_phis
19922 after visiting it, then clear it at the end.
19924 2018-06-01 Richard Biener <rguenther@suse.de>
19926 * tree-vectorizer.h (vect_dr_stmt): New function.
19927 (vect_get_load_cost): Adjust.
19928 (vect_get_store_cost): Likewise.
19929 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
19930 Use vect_dr_stmt instead of DR_SMTT.
19931 (vect_record_base_alignments): Likewise.
19932 (vect_calculate_target_alignment): Likewise.
19933 (vect_compute_data_ref_alignment): Likewise and make static.
19934 (vect_update_misalignment_for_peel): Likewise.
19935 (vect_verify_datarefs_alignment): Likewise.
19936 (vector_alignment_reachable_p): Likewise.
19937 (vect_get_data_access_cost): Likewise. Pass down
19938 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
19939 (vect_get_peeling_costs_all_drs): Likewise.
19940 (vect_peeling_hash_get_lowest_cost): Likewise.
19941 (vect_enhance_data_refs_alignment): Likewise.
19942 (vect_find_same_alignment_drs): Likewise.
19943 (vect_analyze_data_refs_alignment): Likewise.
19944 (vect_analyze_group_access_1): Likewise.
19945 (vect_analyze_group_access): Likewise.
19946 (vect_analyze_data_ref_access): Likewise.
19947 (vect_analyze_data_ref_accesses): Likewise.
19948 (vect_vfa_segment_size): Likewise.
19949 (vect_small_gap_p): Likewise.
19950 (vectorizable_with_step_bound_p): Likewise.
19951 (vect_prune_runtime_alias_test_list): Likewise.
19952 (vect_analyze_data_refs): Likewise.
19953 (vect_supportable_dr_alignment): Likewise.
19954 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
19955 (vect_gen_prolog_loop_niters): Likewise.
19956 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
19957 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
19959 (vect_recog_mask_conversion_pattern): Likewise.
19960 (vect_try_gather_scatter_pattern): Likewise.
19961 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
19962 to vect_get_store_cost.
19963 (vect_get_store_cost): Get stmt_info instead of DR.
19964 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
19965 (vect_get_load_cost): Get stmt_info instead of DR.
19967 2018-06-01 Richard Biener <rguenther@suse.de>
19969 PR middle-end/86017
19970 * gimple-fold.c (var_decl_component_p): Also allow offsetted
19971 vars wrapped in MEM_REFs.
19973 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
19975 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
19976 Fix subreg tests so that we only return a choice between
19977 GENERAL_REGS and FP_REGS if the original classes included both.
19979 2018-06-01 Richard Biener <rguenther@suse.de>
19982 * tree-ssa-structalias.c (get_function_part_constraint):
19983 Handle NULL fi->decl.
19984 (find_func_aliases_for_call): Properly handle indirect
19985 fi from direct call.
19986 (find_func_clobbers): Likewise.
19987 (ipa_pta_execute): Likewise.
19988 (create_variable_info_for): For functions that are ifunc_resolver
19989 resolve to a varinfo that contains the result of the resolver call.
19990 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
19993 2018-05-31 Michael Collison <michael.collison@arm.com>
19995 * config/aarch64/aarch64.md:
19996 (*fix_to_zero_extenddfdi2): New pattern.
19997 * gcc.target/aarch64/fix_extend1.c: New testcase.
19999 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
20001 PR middle-end/78809
20002 PR middle-end/83026
20003 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
20004 and BUILT_IN_STRNCMP_EQ.
20005 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
20006 BUILT_IN_STRNCMP_EQ.
20007 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
20008 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
20009 (gimple_fold_builtin): Likewise.
20010 * tree-ssa-strlen.c (compute_string_length): New function.
20011 (determine_min_obsize): New function.
20012 (handle_builtin_string_cmp): New function to handle calls to
20013 string compare functions.
20014 (strlen_optimize_stmt): Add handling to builtin string compare
20016 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
20017 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
20018 * tree.c (build_common_builtin_nodes): Add new defines of
20019 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
20021 2018-05-31 Jakub Jelinek <jakub@redhat.com>
20024 * bb-reorder.c (pass_partition_blocks::gate): Return false for
20025 functions with naked attribute.
20027 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
20029 * config/i386/sse.md (avx_vec_concat<mode>):
20030 Substitute concat_tg_mode mode attribute with xtg_mode.
20031 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
20032 (concat_tg_mode): Remove mode attribute.
20034 2018-05-31 Martin Sebor <msebor@redhat.com>
20037 * calls.c (alloc_max_size): Correct a logic error/typo.
20038 Treat excessive arguments as infinite. Warn for invalid arguments.
20039 * doc/invoke.texi (-Walloc-size-larger-than): Update.
20041 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
20044 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
20045 and movx for Haswell.
20047 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
20048 Cesar Philippidis <cesar@codesourcery.com>
20050 PR middle-end/85879
20051 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
20052 when emitting error on private/firstprivate reductions.
20053 * omp-low.c (lower_omp_target): Avoid reference-type processing
20054 on pointers for firstprivate clause.
20056 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
20058 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
20061 * config/aarch64/aarch64-simd.md
20062 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
20063 (aarch64_ld1_x3_<mode>): Likewise
20064 (aarch64_st1x2<VALLDIF:mode>): Likewise
20065 (aarch64_st1_x2_<mode>): Likewise
20066 (aarch64_st1x3<VALLDIF:mode>): Likewise
20067 (aarch64_st1_x3_<mode>): Likewise
20068 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
20069 (vld1_s8_x3): Likewise.
20070 (vld1_u16_x3): Likewise.
20071 (vld1_s16_x3): Likewise.
20072 (vld1_u32_x3): Likewise.
20073 (vld1_s32_x3): Likewise.
20074 (vld1_u64_x3): Likewise.
20075 (vld1_s64_x3): Likewise.
20076 (vld1_f16_x3): Likewise.
20077 (vld1_f32_x3): Likewise.
20078 (vld1_f64_x3): Likewise.
20079 (vld1_p8_x3): Likewise.
20080 (vld1_p16_x3): Likewise.
20081 (vld1_p64_x3): Likewise.
20082 (vld1q_u8_x3): Likewise.
20083 (vld1q_s8_x3): Likewise.
20084 (vld1q_u16_x3): Likewise.
20085 (vld1q_s16_x3): Likewise.
20086 (vld1q_u32_x3): Likewise.
20087 (vld1q_s32_x3): Likewise.
20088 (vld1q_u64_x3): Likewise.
20089 (vld1q_s64_x3): Likewise.
20090 (vld1q_f16_x3): Likewise.
20091 (vld1q_f32_x3): Likewise.
20092 (vld1q_f64_x3): Likewise.
20093 (vld1q_p8_x3): Likewise.
20094 (vld1q_p16_x3): Likewise.
20095 (vld1q_p64_x3): Likewise.
20096 (vst1_s64_x2): Likewise.
20097 (vst1_u64_x2): Likewise.
20098 (vst1_f64_x2): Likewise.
20099 (vst1_s8_x2): Likewise.
20100 (vst1_p8_x2): Likewise.
20101 (vst1_s16_x2): Likewise.
20102 (vst1_p16_x2): Likewise.
20103 (vst1_s32_x2): Likewise.
20104 (vst1_u8_x2): Likewise.
20105 (vst1_u16_x2): Likewise.
20106 (vst1_u32_x2): Likewise.
20107 (vst1_f16_x2): Likewise.
20108 (vst1_f32_x2): Likewise.
20109 (vst1_p64_x2): Likewise.
20110 (vst1q_s8_x2): Likewise.
20111 (vst1q_p8_x2): Likewise.
20112 (vst1q_s16_x2): Likewise.
20113 (vst1q_p16_x2): Likewise.
20114 (vst1q_s32_x2): Likewise.
20115 (vst1q_s64_x2): Likewise.
20116 (vst1q_u8_x2): Likewise.
20117 (vst1q_u16_x2): Likewise.
20118 (vst1q_u32_x2): Likewise.
20119 (vst1q_u64_x2): Likewise.
20120 (vst1q_f16_x2): Likewise.
20121 (vst1q_f32_x2): Likewise.
20122 (vst1q_f64_x2): Likewise.
20123 (vst1q_p64_x2): Likewise.
20124 (vst1_s64_x3): Likewise.
20125 (vst1_u64_x3): Likewise.
20126 (vst1_f64_x3): Likewise.
20127 (vst1_s8_x3): Likewise.
20128 (vst1_p8_x3): Likewise.
20129 (vst1_s16_x3): Likewise.
20130 (vst1_p16_x3): Likewise.
20131 (vst1_s32_x3): Likewise.
20132 (vst1_u8_x3): Likewise.
20133 (vst1_u16_x3): Likewise.
20134 (vst1_u32_x3): Likewise.
20135 (vst1_f16_x3): Likewise.
20136 (vst1_f32_x3): Likewise.
20137 (vst1_p64_x3): Likewise.
20138 (vst1q_s8_x3): Likewise.
20139 (vst1q_p8_x3): Likewise.
20140 (vst1q_s16_x3): Likewise.
20141 (vst1q_p16_x3): Likewise.
20142 (vst1q_s32_x3): Likewise.
20143 (vst1q_s64_x3): Likewise.
20144 (vst1q_u8_x3): Likewise.
20145 (vst1q_u16_x3): Likewise.
20146 (vst1q_u32_x3): Likewise.
20147 (vst1q_u64_x3): Likewise.
20148 (vst1q_f16_x3): Likewise.
20149 (vst1q_f32_x3): Likewise.
20150 (vst1q_f64_x3): Likewise.
20151 (vst1q_p64_x3): Likewise.
20153 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
20155 * config/msp430/msp430.c (msp430_output_labelref): Prepend
20156 user_label_prefix to name.
20158 * tree-core.h: Update comment about the format of NAME string
20159 passed to handler in attribute_spec.
20161 * config/msp430/msp430.md: Remove erroneous subreg expression from
20162 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
20163 zero_extend{q,h}isi2.
20165 2018-05-30 Borislav Petkov <bp@suse.de>
20167 * doc/extend.texi: Document some architecture specific
20168 constraints and sort entries.
20170 2018-05-30 Martin Sebor <msebor@redhat.com>
20172 PR middle-end/85369
20173 * builtins.c (expand_builtin_stpcpy_1): New function.
20174 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
20175 only if the former succeeds.
20177 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
20179 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
20182 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
20184 * doc/invoke.texi (-flinker-output): Document
20186 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
20188 * passes.c (ipa_write_summaries): Only modify statements if body
20190 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
20191 incrementally linking.
20192 (ipa_passes): Likewise.
20193 * lto-cgraph.c (lto_output_node): When incrementally linking do not
20194 pass down resolution info.
20195 * common.opt (flag_incremental_link): Update info.
20196 * gcc.c (plugin specs): Turn flinker-output=* to
20197 -plugin-opt=-linker-output-known
20198 * toplev.c (compile_file): Also cut compilation when doing incremental
20200 * flag-types. (enum lto_partition_model): Add
20201 LTO_LINKER_OUTPUT_NOLTOREL.
20202 (invoke.texi): Add -flinker-output docs.
20203 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
20204 link same way as WPA; do not stream in dead initializers.
20206 * dwarf2out.c (dwarf2out_die_ref_for_decl,
20207 darf2out_register_external_decl): Support incremental link.
20209 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
20211 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
20213 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
20215 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
20216 it down to simple_object_copy_lto_debug_sections.
20217 (run_gcc): Determine incremental LTO link time and configure
20218 lto1 into non-wpa mode, disable renaming of debug sections.
20220 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
20222 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
20223 descriptions of various incorrectly documented functions.
20225 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
20228 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
20229 address check not strict.
20231 2018-05-30 Richard Biener <rguenther@suse.de>
20233 PR tree-optimization/85964
20234 * tracer.c (better_p): Drop initialized count check, we only
20235 call the function with initialized counts now.
20236 (find_best_successor): Do find a best edge if one
20237 has uninitialized count.
20238 (find_best_predecessor): Likewise. Do BB frequency check only
20239 if count is initialized.
20241 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
20243 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
20244 (aarch64_ldrstr_offset_compare): New.
20245 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
20246 load/store orderings.
20247 (aarch64_gen_adjusted_ldpstp): Likewise.
20249 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
20251 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
20252 Check for subset of GENERAL_REGS and FP_REGS.
20253 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
20256 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
20258 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
20259 and wi::to_poly_offset. Add the current offset and then check
20260 whether the sum fits, rather than using an unchecked addition of
20261 a checked term. Check for a shwi rather than a uhwi.
20262 * expr.c (get_bit_range): Use tree_to_poly_uint64.
20263 (store_constructor): Use poly_int_tree_p.
20264 (expand_expr_real_1): Likewise.
20265 * function.c (assign_temp): Likewise.
20266 * fold-const.c (const_binop): Use poly_int_tree_p and
20267 wi::to_poly_offset.
20268 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
20270 * ipa-icf-gimple.c (func_checker::compare_operand): Use
20271 to_poly_offset for MEM offsets.
20272 * ipa-icf.c (sem_variable::equals): Likewise.
20273 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
20274 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
20275 wi::to_poly_offset for BIT_FIELD_REF offsets.
20276 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
20277 wi::to_poly_offset.
20278 * var-tracking.c (emit_note_insn_var_location): Use
20279 tree_to_poly_uint64.
20281 2018-05-29 Jim Wilson <jimw@sifive.com>
20283 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
20285 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
20288 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
20289 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
20291 (sse4_1_round<mode>2): Use nonimmediate_operand
20292 for operand 1 predicate.
20294 2018-05-29 Martin Sebor <msebor@redhat.com>
20295 Richard Biener <rguenther@suse.de>
20298 * calls.c (get_size_range): Call determine_value_range instead
20299 of get_value_range..
20300 * tree-vrp.h (determine_value_range): Declared new function.
20301 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
20303 2018-05-29 Richard Biener <rguenther@suse.de>
20305 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
20306 sure to use non-pattern stmts for get_earlier_stmt arguments.
20307 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
20308 called on pattern stmts.
20309 (get_later_stmt): Likewise.
20311 2018-05-29 Martin Liska <mliska@suse.cz>
20313 PR gcov-profile/85759
20314 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
20317 2018-05-29 Jakub Jelinek <jakub@redhat.com>
20319 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
20321 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
20325 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
20326 VEC_PACK_FLOAT_EXPR): New tree codes.
20327 * tree-pretty-print.c (op_code_prio): Handle
20328 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
20329 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
20330 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
20331 * tree-inline.c (estimate_operator_cost): Likewise.
20332 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
20333 * fold-const.c (const_binop): Likewise.
20334 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
20335 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
20336 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
20337 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
20338 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
20339 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
20340 * expr.c (expand_expr_real_2): Likewise.
20341 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
20342 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
20343 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
20345 * optabs.c (expand_widen_pattern_expr): For
20346 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
20347 sign from result type rather than operand's type.
20348 (expand_binop_directly): For vec_packu_float_optab and
20349 vec_packs_float_optab allow result type to be different from operand's
20351 * optabs-tree.c (optab_for_tree_code): Handle
20352 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
20353 VEC_PACK_FLOAT_EXPR. Formatting fixes.
20354 * tree-vect-generic.c (expand_vector_operations_1): Handle
20355 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
20356 VEC_PACK_FLOAT_EXPR.
20357 * tree-vect-stmts.c (supportable_widening_operation): Handle
20359 (supportable_narrowing_operation): Handle FLOAT_EXPR.
20360 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
20361 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
20362 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
20363 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
20365 (vec_pack<floatprefix>_float_<mode>): New expander.
20366 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
20368 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
20369 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
20370 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
20371 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
20372 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
20374 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
20375 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
20376 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
20377 VEC_PACK_FLOAT_EXPR): Document.
20379 2018-05-29 Richard Biener <rguenther@suse.de>
20381 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
20383 (stmt_vec_info_vec): Make pointer.
20384 (init_stmt_vec_info_vec): Remove.
20385 (free_stmt_vec_info_vec): Likewise.
20386 (set_stmt_vec_info_vec): New function.
20387 (free_stmt_vec_infos): Likewise.
20388 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
20389 (set_vinfo_for_stmt): Likewise.
20390 (get_earlier_stmt): Likewise.
20391 (get_later_stmt): Likewise.
20392 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
20393 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
20394 (vec_info::~vec_info): Free stmt_vec_infos.
20395 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
20396 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
20397 (pass_slp_vectorize::execute): Likewise.
20398 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
20399 (free_stmt_vec_info_vec): Likewise.
20400 (set_stmt_vec_info_vec): New function.
20401 (free_stmt_vec_infos): Likewise.
20402 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
20403 the global stmt_vec_info_vec.
20404 * tree-parloops.c (gather_scalar_reductions): Use
20405 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
20408 2018-05-29 Richard Biener <rguenther@suse.de>
20410 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
20412 2018-05-29 Martin Liska <mliska@suse.cz>
20413 David Malcolm <dmalcolm@redhat.com>
20415 * vec.c (test_reverse): New.
20416 (vec_c_tests): Add new test.
20417 * vec.h (vl_ptr>::reverse): New function.
20419 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
20421 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
20423 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
20426 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
20428 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
20430 2018-05-28 Richard Biener <rguenther@suse.de>
20432 PR tree-optimization/85933
20433 * tree-vect-data-refs.c (vect_record_base_alignments): Only
20434 look at stmts marked as vectorizable.
20436 2018-05-28 Richard Biener <rguenther@suse.de>
20438 PR tree-optimization/85934
20439 * tree-vect-generic.c (expand_vector_operations_1): Hoist
20440 vector boolean check before scalar optimization.
20442 2018-05-28 Jakub Jelinek <jakub@redhat.com>
20444 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
20447 2018-05-28 Mark Wielaard <mark@klomp.org>
20449 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
20450 if it is an expression containing a minus sign.
20452 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
20454 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
20456 2018-05-27 Paul Koning <ni1d@arrl.net>
20458 * config/pdp11/pdp11.md (truncsihi2): Remove.
20460 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
20461 Chung-Ju Wu <jasonwucj@gmail.com>
20463 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
20465 (unaligned_store_dw): Ditto.
20466 * config/nds32/nds32-memory-manipulation.c
20467 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
20468 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
20469 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
20470 (emit_setmem_word_loop): Rename to ...
20471 (emit_setmem_doubleword_loop): ... this.
20472 (nds32_gen_dup_4_byte_to_word_value): New function.
20473 (nds32_gen_dup_8_byte_to_double_word_value): New function.
20474 (nds32_expand_setmem_loop): Refine implementation.
20475 (nds32_expand_setmem_loop_v3m): Ditto.
20476 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
20479 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
20481 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
20483 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
20485 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
20486 (nds32_init_machine_status): Initialize machine->attr_naked_p and
20487 machine->attr_no_prologue_p.
20488 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
20489 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
20490 (nds32_expand_epilogue): Consider attr_naked_p.
20491 (nds32_expand_epilogue_v3pop): Likewise.
20492 (nds32_can_use_return_insn): Likewise.
20493 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
20494 attr_no_prologue_p fields.
20495 * config/nds32/nds32.opt (mret-in-naked-func): New option.
20497 2018-05-27 Jakub Jelinek <jakub@redhat.com>
20500 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
20502 * config/i386/sse.md
20503 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
20505 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
20507 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
20509 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
20511 (*<floatsuffix>floatv2div2sf2): Rename to ...
20512 (*float<floatunssuffix>v2div2sf2): ... this.
20513 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
20514 (float<floatunssuffix>v2div2sf2_mask): ... this.
20515 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
20516 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
20517 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
20519 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
20521 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
20523 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
20525 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
20527 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
20529 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
20530 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
20531 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
20532 gen_ufix_truncv8dfv8si2.
20533 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
20534 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
20535 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
20536 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
20537 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
20538 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
20539 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
20540 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
20542 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
20546 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
20548 2018-05-25 Jim Wilson <jimw@sifive.com>
20550 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
20551 * config/riscv/riscv.c (struct machine_function): Add
20552 interrupt_handler_p and attribute_checked_p fields.
20553 (riscv_attribute_table): Add interrupt.
20554 (riscv_interrupt_type_p): New.
20555 (riscv_save_reg_p): Save extra regs for interrupt handler.
20556 (riscv_use_save_libcall): Return false for interrupt handler.
20557 (riscv_first_stack_step): Add forward declaration.
20558 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
20559 for interrupt handler with large frame. Use it for saved reg list.
20560 (riscv_expand_prologue): Move flag_stack_usage_info support to
20561 eliminate duplication.
20562 (riscv_expand_epilogue): Generate mret for interrupt handler.
20563 (riscv_epilogue_uses): New.
20564 (riscv_can_use_return_insn): Return false for interrupt handler.
20565 (riscv_function_ok_for_sibcall): Likewise.
20566 (riscv_set_current_function): Add interrupt handler support.
20567 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
20568 * config/riscv/riscv.md (UNSPECV_MRET): New.
20570 (riscv_frflags, riscv_fsflags): Use tab after opcode.
20572 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
20574 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
20576 PR tree-optimization/85712
20577 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
20578 this candidate has already been replaced in-situ by a copy.
20580 2018-05-25 Jason Merrill <jason@redhat.com>
20582 PR c++/80485 - inline function non-zero address.
20583 * symtab.c (nonzero_address): Check DECL_COMDAT.
20585 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
20588 * config/alpha/alpha.md (ashlsi3): New insn pattern.
20589 (*ashlsi_se): Rename from *ashldi_se. Define as sign
20590 extension of SImode operation. Use const123_operand predicate.
20591 (*saddsi_1): Remove.
20592 (*saddl_se_1): Ditto.
20593 (*ssubsi_1): Ditto.
20594 (*ssubl_se_1): Ditto.
20595 * config/alpha/predicates.md (const123_operand): New predicate.
20596 * config/alpha/constraints.md (P): Use IN_RANGE.
20598 2018-05-25 Richard Biener <rguenther@suse.de>
20600 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
20602 (ref_maybe_used_by_stmt_p): Likewise.
20603 (stmt_may_clobber_ref_p): Likewise.
20604 (stmt_may_clobber_ref_p_1): Likewise.
20605 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
20607 (ref_maybe_used_by_stmt_p): Likewise.
20608 (stmt_may_clobber_ref_p): Likewise.
20609 (stmt_may_clobber_ref_p_1): Likewise.
20610 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
20611 the alias oracle to disambiguate DRs with stmts DR analysis
20613 (vect_analyze_data_refs): Do not give up on not analyzable
20614 DRs for BB vectorization. Remove code truncating the dataref
20617 2018-05-25 Jakub Jelinek <jakub@redhat.com>
20620 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
20621 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
20622 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
20624 2018-05-25 Richard Biener <rguenther@suse.de>
20626 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
20627 function, combining stmt data ref gathering and fatal analysis
20629 (vect_analyze_data_refs): Remove now redudnant code and simplify.
20630 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
20631 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
20632 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
20633 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
20635 2018-05-25 Bin Cheng <bin.cheng@arm.com>
20637 PR tree-optimization/85720
20638 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
20639 SCC if all partitions are builtins.
20640 (version_loop_by_alias_check): New parameter. Generate cancelable
20641 runtime alias check if all partitions are builtins.
20642 (distribute_loop): Update call to above function.
20644 2018-05-25 Bin Cheng <bin.cheng@arm.com>
20646 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
20647 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
20648 (parm_default_def_partition_arg): Ditto.
20649 (set_parm_default_def_partition): Ditto.
20650 (get_parm_default_def_partitions): Ditto and make it static.
20651 (get_undefined_value_partitions): Ditto and make it static.
20652 (remove_ssa_form): Refactor call to init_var_map here.
20653 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
20654 computation for loop region.
20655 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
20656 (register_default_def): Delete.
20657 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
20658 (parm_default_def_partition_arg): Ditto.
20659 (set_parm_default_def_partition): Ditto.
20660 (get_parm_default_def_partitions): Ditto and make it static.
20661 (get_undefined_value_partitions): Ditto and make it static.
20662 (coalesce_with_default, coalesce_with_default): Update comment.
20663 (create_coalesce_list_for_region): New func factored out from
20664 create_outofssa_var_map.
20665 (populate_coalesce_list_for_outofssa): New func factored out from
20666 create_outofssa_var_map and coalesce_ssa_name.
20667 (create_outofssa_var_map): Delete.
20668 (coalesce_ssa_name): Refactor to support live range computation.
20669 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
20670 (get_parm_default_def_partitions): Delete.
20671 (get_undefined_value_partitions): Ditto.
20672 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
20673 computation for loop region.
20674 (new_tree_live_info, loe_visit_block): Ditto.
20675 (live_worklist, set_var_live_on_entry): Ditto.
20676 (calculate_live_on_exit, verify_live_on_entry): Ditto.
20677 * tree-ssa-live.h (struct _var_map): New fields.
20678 (init_var_map): Change decl.
20679 (region_contains_p): New.
20681 2018-05-25 Bin Cheng <bin.cheng@arm.com>
20683 * tree-ssa-live.h (live_merge_and_clear): Delete.
20685 2018-05-25 Richard Biener <rguenther@suse.de>
20688 * tree-dump.c (dequeue_and_dump): Remove access to removed
20689 operand 2 of a SWITCH_EXPR.
20691 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20693 * doc/sourcebuild.texi (vect_double_cond_arith): Include
20694 multiplication and division.
20695 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
20696 (cond_udiv@var{m}, cond_umod@var{m}): Document.
20697 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
20698 (cond_udiv_optab, cond_umod_optab): New optabs.
20699 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
20700 (IFN_COND_RDIV): New internal functions.
20701 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
20702 TRUNC_MOD_EXPR and RDIV_EXPR.
20703 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
20704 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
20706 (SVE_INT_BINARY): Include mult.
20707 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
20708 (optab, sve_int_op): Handle mult.
20709 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
20711 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
20712 for SVE_INT_BINARY_SD.
20714 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20716 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
20717 (optab, sve_int_op): Handle div and udiv.
20718 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
20719 for SVE_INT_BINARY_SD.
20720 (*<optab><mode>3): New insn for the same.
20722 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20724 * tree-vect-patterns.c: Include predict.h.
20725 (vect_recog_divmod_pattern): Restrict check for division support
20726 to when optimizing for size.
20728 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20730 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
20731 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
20732 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
20733 (gimple_match_op::set_op): Likewise.
20734 (gimple_resimplify4): Declare.
20735 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
20736 (expr::gen_transform): Likewise.
20737 (decision_tree::gen): Generate a simplification routine for 4 operands.
20738 * gimple-match-head.c (gimple_simplify): Add an overload for
20739 4 operands. In the top-level function, handle up to 4 call
20740 arguments and call gimple_resimplify4.
20741 (gimple_resimplify4): New function.
20742 (build_call_internal): Pass a fourth operand.
20743 (maybe_push_to_seq): Likewise.
20744 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
20745 Fold VEC_COND_EXPRs of an operation and a default value into
20746 an IFN_COND_* function if possible.
20747 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
20749 (SVE_COND_FP_BINARY): Include them.
20750 (optab, sve_fp_op): Handle them.
20751 (SVE_INT_BINARY_REV): New code iterator.
20752 (SVE_COND_FP_BINARY_REV): New int iterator.
20753 (commutative): New int attribute.
20754 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
20756 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
20758 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
20759 (*cond_<optab><mode>): New patterns for reversed operands.
20761 2018-05-25 Richard Biener <rguenther@suse.de>
20763 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
20764 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
20765 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
20766 (STMT_VINFO_GROUPED_ACCESS): Adjust.
20767 * tree-vect-data-refs.c (everywhere): Adjust users.
20768 * tree-vect-loop.c (everywhere): Likewise.
20769 * tree-vect-slp.c (everywhere): Likewise.
20770 * tree-vect-stmts.c (everywhere): Likewise.
20771 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
20773 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20775 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
20777 (gcc_cv_as_section_exclude): ... this.
20778 Try Solaris as #exclude syntax.
20779 * configure: Regenerate.
20780 * config.in: Regenerate.
20781 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
20783 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
20784 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
20786 * varasm.c (default_elf_asm_named_section): Don't check if
20787 HAVE_GAS_SECTION_EXCLUDE is defined.
20789 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20791 * doc/md.texi: Update the documentation of the cond_* optabs
20792 to mention the new final operand. Fix GET_MODE_NUNITS call.
20793 Describe the scalar case too.
20794 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
20795 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
20797 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
20798 (get_conditional_internal_fn): Update comment.
20799 * tree-vect-loop.c (vectorizable_reduction): Pass the original
20800 accumulator value as a final argument to conditional functions.
20801 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
20802 a define_expand and add an "else" operand. Assert for now that
20803 the else operand is equal to operand 2. Use SVE_INT_BINARY and
20804 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
20805 (*cond_<optab><mode>): New patterns.
20806 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
20807 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
20808 (UNSPEC_COND_EOR): Delete.
20809 (optab): Remove associated mappings.
20810 (SVE_INT_BINARY): New code iterator.
20811 (sve_int_op): Remove int attribute and add "minus" to the code
20813 (SVE_COND_INT_OP): Delete.
20814 (SVE_COND_FP_OP): Rename to...
20815 (SVE_COND_FP_BINARY): ...this.
20817 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20819 * optabs.c (can_reuse_operands_p): New function.
20820 (maybe_legitimize_operands): Try to reuse the results for
20823 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
20825 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
20826 Add {q} suffix to insn mnemonic.
20828 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
20830 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
20831 (msp430_warn_func_return): New.
20833 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
20835 * fold-const.c (tree_nonzero_bits): New function.
20836 * fold-const.h (tree_nonzero_bits): Likewise.
20837 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
20838 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
20840 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
20844 * varasm.c (assemble_alias): Check ifunc_resolver only on
20847 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
20850 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
20851 when memory input operand is handled.
20853 2018-05-24 Luis Machado <luis.machado@linaro.org>
20855 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
20857 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
20859 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
20861 * match.pd: Delay FMA folds until after vectorization.
20863 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
20866 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
20867 address check not strict.
20869 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
20871 * gimple-match.h (gimple_match_op): New class.
20872 (mprts_hook): Replace parameters with a gimple_match_op *.
20873 (maybe_build_generic_op): Likewise.
20874 (gimple_simplified_result_is_gimple_val): Replace parameters with
20875 a const gimple_match_op *.
20876 (gimple_simplify): Replace code_helper * and tree * parameters with
20877 a gimple_match_op * parameter.
20878 (gimple_resimplify1): Replace code_helper *, tree and tree *
20879 parameters with a gimple_match_op * parameter.
20880 (gimple_resimplify2): Likewise.
20881 (gimple_resimplify3): Likewise.
20882 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
20883 parameters with a gimple_match_op * parameter.
20884 * gimple-match-head.c (gimple_simplify): Change prototypes of
20885 auto-generated functions to take a gimple_match_op * instead of
20886 separate code_helper * and tree * parameters. Make the same
20887 change in the top-level overload and update calls to the
20888 gimple_resimplify routines. Update calls to the auto-generated
20889 functions and to maybe_push_res_to_seq in the publicly-facing
20890 operation-specific gimple_simplify overloads.
20891 (gimple_match_op::MAX_NUM_OPS): Define.
20892 (gimple_resimplify1): Replace rcode and ops with a single res_op
20893 parameter. Update call to gimple_simplify.
20894 (gimple_resimplify2): Likewise.
20895 (gimple_resimplify3): Likewise.
20896 (mprts_hook): Replace parameters with a gimple_match_op *.
20897 (maybe_build_generic_op): Likewise.
20898 (build_call_internal): Replace type, nargs and ops with
20899 a gimple_match_op *.
20900 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
20901 with a single gimple_match_op *. Update calls to mprts_hook,
20902 build_call_internal and gimple_simplified_result_is_gimple_val.
20903 Factor out code that is common to the tree_code and combined_fn cases.
20904 * genmatch.c (expr::gen_transform): Replace tem_code and
20905 tem_ops with a gimple_match_op called tem_op. Update calls
20906 to the gimple_resimplify functions and maybe_push_res_to_seq.
20907 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
20908 res_ops. Update call to the gimple_resimplify functions.
20909 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
20910 (decision_tree::gen): Make the functions take a gimple_match_op *
20911 called res_op instead of separate res_code and res_ops parameters.
20912 Update call accordingly.
20913 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
20914 and ops with a single res_op parameter. Update calls to
20915 maybe_build_generic_op and maybe_push_res_to_seq.
20916 (fold_stmt_1): Update calls to gimple_simplify and
20917 replace_stmt_with_simplification.
20918 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
20919 and gimple_simplified_result_is_gimple_val.
20920 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
20922 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
20923 with a gimple_match_op *.
20924 (vn_nary_build_or_lookup): Likewise. Update call to
20925 vn_nary_build_or_lookup_1.
20926 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
20927 gimple_match_op *. Update calls to the gimple_resimplify routines
20928 and to gimple_simplified_result_is_gimple_val.
20929 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
20930 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
20931 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
20932 (visit_nary_op): Likewise.
20933 (visit_reference_op_load): Likewise.
20935 2018-05-23 Luis Machado <luis.machado@linaro.org>
20937 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
20938 modifier for printing the step amount.
20940 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20943 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
20946 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
20948 * doc/sourcebuild.texi (Endianness): New subsubsection.
20950 2018-05-23 Luis Machado <luis.machado@linaro.org>
20952 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20953 <prefetch_dynamic_strides>: New const bool field.
20954 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20955 prefetch_dynamic_strides.
20956 (exynosm1_prefetch_tune): Likewise.
20957 (thunderxt88_prefetch_tune): Likewise.
20958 (thunderx_prefetch_tune): Likewise.
20959 (thunderx2t99_prefetch_tune): Likewise.
20960 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
20962 (aarch64_override_options_internal): Update to set
20963 PARAM_PREFETCH_DYNAMIC_STRIDES.
20964 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
20965 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
20966 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
20967 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
20968 prefetch-dynamic-strides setting.
20970 2018-05-23 Luis Machado <luis.machado@linaro.org>
20972 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20973 <minimum_stride>: New const int field.
20974 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20975 minimum_stride field defaulting to -1.
20976 (exynosm1_prefetch_tune): Likewise.
20977 (thunderxt88_prefetch_tune): Likewise.
20978 (thunderx_prefetch_tune): Likewise.
20979 (thunderx2t99_prefetch_tune): Likewise.
20980 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
20981 <default_opt_level>: Set to 3.
20982 (aarch64_override_options_internal): Update to set
20983 PARAM_PREFETCH_MINIMUM_STRIDE.
20984 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
20985 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
20986 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
20987 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
20988 stride is constant and is below the minimum stride threshold.
20990 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20992 * config/arm/arm-cpus.in (mode26): Delete.
20993 (armv4): Delete mode26 reference.
20994 * config/arm/arm.c (arm_configure_build_target): Delete use of
20997 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
20999 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
21001 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
21002 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
21004 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
21005 (floatunsdidf2): Ditto.
21007 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
21009 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
21010 (fixuns_trunc<mode>si2_avx512f): Ditto.
21011 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
21012 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
21013 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
21015 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
21017 PR rtl-optimization/79985
21018 * df-scan.c (df_insn_refs_collect): Remove special case for
21019 global registers and asm statements.
21021 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
21023 * extend.texi (Global Register Variables): Rewrite the bullet list.
21024 Note that the register is available for allocation. Note that access
21025 via inline asm must use constraints. Add note about async-signal
21026 handlers. Remove paragraph about automagic register selection.
21028 2018-05-23 Richard Biener <rguenther@suse.de>
21030 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
21031 of fixed offset from memset VN.
21033 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
21035 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
21036 first_interp field.
21037 (alloc_cand_and_find_basis): Initialize first_interp field.
21038 (slsr_process_mul): Modify first_interp field.
21039 (slsr_process_add): Likewise.
21040 (slsr_process_cast): Modify first_interp field for each new
21042 (slsr_process_copy): Likewise.
21043 (dump_candidate): Dump first_interp field.
21044 (replace_mult_candidate): Process all interpretations, not just
21046 (replace_rhs_if_not_dup): Likewise.
21047 (replace_one_candidate): Likewise.
21049 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
21051 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
21053 (aarch64_needs_frame_chain): New function.
21054 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
21056 2018-05-23 Sudakshina Das <sudi.das@arm.com>
21059 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
21060 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
21061 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
21062 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
21063 as true for strict-align.
21064 (aarch64_can_inline_p): Perform checks even when callee has no
21065 attributes to check for strict alignment.
21066 * doc/extend.texi (AArch64 Function Attributes): Document
21068 * doc/invoke.texi: (AArch64 Options): Likewise.
21070 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
21072 PR tree-optimization/85853
21073 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
21074 the handling of the root of the node to...
21075 (vect_slp_analyze_node_operations_1): ...this new function,
21076 and run the whole thing with the child nodes' def types
21077 set according to their SLP node's def type.
21079 2018-05-23 Richard Biener <rguenther@suse.de>
21081 PR middle-end/85874
21082 * tree-data-ref.c (create_runtime_alias_checks): Defer
21083 and ignore overflow warnings.
21085 2018-05-23 Yury Gribov <tetra2005@gmail.com>
21087 PR tree-optimization/85822
21088 * tree-vrp.c (is_masked_range_test): Fix handling of negative
21091 2018-05-23 Richard Biener <rguenther@suse.de>
21093 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
21094 memset constants via native_interpret_expr.
21096 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
21099 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
21101 (cgraph_node::create_alias): Likewise.
21102 (cgraph_node::get_availability): Check ifunc_resolver instead
21103 of looking up ifunc attribute.
21104 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
21105 * varasm.c (do_assemble_alias): Likewise.
21106 (assemble_alias): Likewise.
21107 (default_binds_local_p_3): Likewise.
21108 * cgraph.h (cgraph_node): Add ifunc_resolver.
21109 (cgraph_node::only_called_directly_or_aliased_p): Return false
21110 for IFUNC resolver.
21111 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
21113 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
21114 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
21115 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
21116 instead of looking up ifunc attribute.
21118 2018-05-22 Luis Machado <luis.machado@linaro.org>
21120 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
21122 2018-05-22 Martin Sebor <msebor@redhat.com>
21124 PR middle-end/85359
21125 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
21126 only when expasion succeeds.
21127 (expand_builtin_strcmp): Same.
21128 (expand_builtin_strncmp): Same.
21130 2018-05-22 Martin Sebor <msebor@redhat.com>
21132 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
21134 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
21135 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21137 * config/aarch64/aarch64-ldpstp.md: Replace uses of
21138 aarch64_mem_pair_operand with memory_operand and delete operand swapping
21140 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
21141 Add check for legitimate_address.
21142 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
21143 (aarch64_swap_ldrstr_operands): New.
21144 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
21147 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
21148 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21150 * config/aarch64/aarch64.md: New patterns to generate stp
21152 (store_pair_sw, store_pair_dw): New patterns to generate stp for
21153 single words and double words.
21154 (load_pair_sw, load_pair_dw): Likewise.
21155 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
21157 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
21159 * config/aarch64/aarch64-ldpstp.md: Modify peephole
21160 for different mode ldpstp and add peephole for merged zero stores.
21161 Likewise for loads.
21162 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
21164 (aarch64_gen_store_pair): Rename calls to match new patterns.
21165 (aarch64_gen_load_pair): Rename calls to match new patterns.
21166 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
21167 (load_pair<DREG:mode><DREG2:mode>): ... This.
21168 (store_pair<mode>): Rename to...
21169 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
21170 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
21171 New mode iterators.
21172 (V_INT_EQUIV): Handle SImode.
21173 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
21176 2018-05-22 Martin Sebor <msebor@redhat.com>
21179 * calls.c (maybe_warn_nonstring_arg): Use string length to set
21180 or ajust the presumed bound on an operation to avoid unnecessary
21183 2018-05-22 Martin Sebor <msebor@redhat.com>
21185 PR tree-optimization/85826
21186 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
21187 assuming that a DECL necesarily has a constant size.
21189 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
21191 PR middle-end/85862
21192 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
21194 2018-05-22 Richard Biener <rguenther@suse.de>
21196 PR tree-optimization/85834
21197 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
21198 non-constant and non-zero memset arguments.
21200 2018-05-22 Martin Liska <mliska@suse.cz>
21203 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
21205 2018-05-22 Richard Biener <rguenther@suse.de>
21207 PR tree-optimization/85863
21208 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
21209 comparisons when vectype is specified.
21210 (vectorizable_condition): Do not specify vectype for
21211 vect_is_simple_cond when SLP vectorizing.
21213 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
21216 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
21217 define __ibm128 as long double.
21218 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
21219 as a distinct type when IEEE 128-bit support is enabled.
21220 (init_float128_ieee): Fix up conversions between IFmode and IEEE
21221 128-bit types to use the correct functions.
21222 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
21223 convert between 128-bit floating point types that have different
21224 modes but the same representation, instead of using gen_lowpart to
21226 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
21228 (IFKF_reg): New attributes to give the register constraints for
21230 (extend<mode>tf2_internal): New insns to mark an explicit
21231 conversion between 128-bit floating point types that have a
21232 different mode but share the same representation.
21234 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
21236 PR tree-optimization/85814
21237 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
21238 a null return from get_strinfo when unsharing the next
21239 strinfo in the chain.
21241 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
21244 * varasm.c (weak_finish): Clean up weak_decls.
21246 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21248 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
21249 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
21250 UNSPEC_UADALP values.
21251 * config/aarch64/iterators.md (ABAL): New int iterator.
21254 (sur): Add mappings for the above.
21255 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
21257 (aarch64_<sur>abal<mode>_4): Likewise.
21258 (aarch64_<sur>adalp<mode>_3): Likewise.
21259 (<sur>sadv16qi): New define_expand.
21261 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
21263 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
21264 (*movdf_internal): Ditto.
21265 (*rcpsf2_sse): Ditto.
21266 (*rsqrtsf2_sse): Ditto.
21267 (*sqrt<mode>2_sse): Ditto.
21269 2018-05-21 Tamar Christina <tamar.christina@arm.com>
21271 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
21273 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
21274 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
21275 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
21276 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
21278 * config/aarch64/arm_neon.h: Likewise.
21279 * config/aarch64/iterators.md (VQ_I): New.
21281 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
21283 * config.gcc: Add arc/t-multilib-linux to tmake_file for
21285 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
21288 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
21290 * config/nds32/constraints.md (S): New constraint.
21291 * config/nds32/nds32.md (call_internal): Use constraint S.
21292 (call_value_internal): Likewise.
21293 (sibcall_internal): Likewise.
21294 (sibcall_value_internal): Likewise.
21296 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
21297 Chung-Ju Wu <jasonwucj@gmail.com>
21299 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
21300 into consideration.
21302 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
21303 Chung-Ju Wu <jasonwucj@gmail.com>
21305 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
21306 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
21307 (nds32_rtx_costs_impl): Simplify.
21308 (nds32_address_cost_impl): Simplify.
21309 (nds32_init_rtx_costs): New function.
21310 (nds32_rtx_costs_speed_prefer): Likewise.
21311 (nds32_rtx_costs_size_prefer): Likewise.
21312 (nds32_address_cost_speed_prefer): Likewise.
21313 (nds32_address_cost_speed_fwprop): Likewise.
21314 (nds32_address_cost_size_prefer): Likewise.
21315 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
21316 * config/nds32/nds32.c (nds32_option_override): Use
21317 nds32_init_rtx_costs function.
21319 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
21321 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
21322 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
21323 (TARGET_PIPELINE_N8): Likewise.
21324 (TARGET_PIPELINE_N10): Likewise.
21325 (TARGET_PIPELINE_N13): Likewise.
21326 (TARGET_PIPELINE_GRAYWOLF): Likewise.
21328 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
21330 * config/nds32/nds32-fpu.md: Update copyright year.
21332 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21334 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
21336 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21338 * config/nds32/nds32.c
21339 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
21340 * config/nds32/nds32.opt (minline-asm-r15): New option.
21342 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21344 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
21346 * config/nds32/nds32.md (abssi2): New pattern.
21348 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
21350 * config/i386/i386.md (rex64namesuffix): New mode attribute.
21351 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
21352 Merge insn pattern from sse_cvtsi2ss<round_name> and
21353 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
21354 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
21355 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
21356 using SWI48 mode iterator.
21357 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
21358 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
21359 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
21360 pattern from sse_cvttss2si<round_saeonly_name>
21361 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
21362 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
21363 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
21364 using SWI48 mode iterator.
21365 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
21366 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
21367 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
21368 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
21369 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
21370 using SWI48 mode iterator.
21371 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
21372 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
21373 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
21374 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
21375 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
21376 SWI48 mode iterator.
21377 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
21378 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
21379 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
21380 pattern from sse_cvttsd2si<round_saeonly_name>
21381 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
21383 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21385 * config/nds32/nds32-md-auxiliary.c
21386 (nds32_valid_smw_lwm_base_p): Refine.
21387 (nds32_output_smw_single_word): Refine.
21388 (nds32_output_smw_double_word): New.
21389 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
21391 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21393 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
21394 (nds32_output_stack_pop): Refine.
21395 (nds32_expand_unaligned_load): Refine.
21396 (nds32_expand_unaligned_store): Refine.
21398 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
21399 Chung-Ju Wu <jasonwucj@gmail.com>
21401 * config/nds32/constants.md: Add TP_REGNUM constant.
21402 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
21403 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
21405 * config/nds32/nds32-doubleword.md: Consider flag_pic.
21406 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
21407 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
21408 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
21409 and PIC code generation.
21410 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
21412 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
21414 * config/nds32/nds32.md: Support TLS and PIC.
21415 * config/nds32/nds32.c: Support TLS and PIC.
21416 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
21417 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
21420 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21422 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
21423 mode with E_ prefix.
21425 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
21426 Chung-Ju Wu <jasonwucj@gmail.com>
21428 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
21429 * config/nds32/nds32-md-auxiliary.c
21430 (symbolic_reference_mentioned_p): New.
21431 (nds32_legitimize_ict_address): New.
21432 (nds32_expand_ict_move): New.
21433 (nds32_indirect_call_referenced_p): New.
21434 (nds32_symbol_binds_local_p): Delete.
21435 (nds32_long_call_p): Modify.
21436 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
21437 * config/nds32/nds32-protos.h
21438 (symbolic_reference_mentioned_p): Declare.
21439 (nds32_legitimize_ict_address): Declare.
21440 (nds32_expand_ict_move): Declare.
21441 (nds32_indirect_call_referenced_p): Declare.
21442 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
21443 (nds32_relax_group): Use nds32_ict_const_p as condition.
21444 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
21445 (nds32_asm_file_start): Output ict_model directive in asm code.
21446 (nds32_legitimate_address_p): Consider indirect call.
21447 (nds32_print_operand): Consider indirect call.
21448 (nds32_print_operand_address): Consider indirect call.
21449 (nds32_insert_attributes): Handle "indirect_call" attribute.
21450 (TARGET_LEGITIMATE_ADDRESS_P): Define.
21451 (TARGET_LEGITIMATE_CONSTANT_P): Define.
21452 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
21453 (TARGET_DELEGITIMIZE_ADDRESS): Define.
21454 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
21455 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
21456 (TARGET_ICT_MODEL_SMALL): Define.
21457 (TARGET_ICT_MODEL_LARGE): Define.
21458 * config/nds32/nds32.md (movsi): Consider ict model.
21459 (call, call_value): Consider ict model.
21460 (sibcall, sibcall_value): Consider ict model.
21461 * config/nds32/nds32.opt (mict-model): New option.
21462 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
21465 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
21466 Monk Chiang <sh.chiang04@gmail.com>
21467 Jim Wilson <jimw@sifive.com>
21469 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
21470 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
21471 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
21472 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
21473 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
21474 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
21475 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
21476 compute save_libcall_adjustment properly.
21477 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
21478 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
21479 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
21480 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
21481 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
21482 (ABI_SPEC): Handle mabi=ilp32e.
21483 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
21484 (RVE): Add RVE mask.
21485 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
21486 <-march>: Add rv32e as an example.
21488 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
21491 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
21492 (intra_create_variable_infos): Handle C++ constructors.
21494 2018-05-18 Martin Liska <mliska@suse.cz>
21496 * passes.def: Remove a redundant pass.
21498 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
21501 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
21503 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21505 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
21507 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
21509 (armv2, armv2a, armv3, armv3m): Delete architectures.
21510 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
21511 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
21512 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
21514 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
21515 (*mulsidi3adddi): Likewise.
21516 (mulsidi3): Likewise.
21517 (*mulsidi3_nov6): Likewise.
21518 (umulsidi3): Likewise.
21519 (umulsidi3_nov6): Likewise.
21520 (umaddsidi4): Likewise.
21521 (*umulsidi3adddi): Likewise.
21522 (smulsi3_highpart): Likewise.
21523 (*smulsi3_highpart_nov6): Likewise.
21524 (umulsi3_highpart): Likewise.
21525 (*umulsi3_highpart_nov6): Likewise.
21526 * config/arm/arm.h (arm_arch3m): Delete.
21527 * config/arm/arm.c (arm_arch3m): Delete.
21528 (arm_option_override_internal): Update armv3-related comment.
21529 (arm_configure_build_target): Delete use of isa_bit_mode32.
21530 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
21531 (arm_rtx_costs_internal): Delete check of arm_arch3m.
21532 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
21533 (mulsa3): Likewise.
21534 (mulusa3): Likewise.
21535 * config/arm/arm-protos.h (arm_arch3m): Delete.
21536 * config/arm/arm-tables.opt: Regenerate.
21537 * config/arm/arm-tune.md: Likewise.
21538 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
21539 deleted architectures.
21541 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21543 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
21544 (armv5t, armv5te): New features.
21545 (ARMv5, ARMv5e): Delete fgroups.
21546 (ARMv5t, ARMv5te): Adjust for above changes.
21547 (ARMv6m): Likewise.
21548 (armv5, armv5e): Delete arches.
21549 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
21551 (*call_reg_arm): Likewise.
21552 (*call_value_reg_armv5): Likewise.
21553 (*call_value_reg_arm): Likewise.
21554 (*call_symbol): Likewise.
21555 (*call_value_symbol): Likewise.
21556 (*sibcall_insn): Likewise.
21557 (*sibcall_value_insn): Likewise.
21558 (clzsi2): Likewise.
21559 (prefetch): Likewise.
21560 (define_split and define_peephole2 dependent on arm_arch5):
21562 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
21564 (TARGET_ARM_QBIT): Likewise.
21565 (TARGET_DSP_MULTIPLY): Likewise.
21566 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
21567 (arm_arch5, arm_arch5e): Delete.
21568 (arm_arch5t, arm_arch5te): Declare.
21569 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
21570 (arm_arch5t): Declare.
21571 (arm_option_reconfigure_globals): Update for the above.
21572 (arm_options_perform_arch_sanity_checks): Update comment, replace
21573 use of arm_arch5 with arm_arch5t.
21574 (use_return_insn): Likewise.
21575 (arm_emit_call_insn): Likewise.
21576 (output_return_instruction): Likewise.
21577 (arm_final_prescan_insn): Likewise.
21578 (arm_coproc_builtin_available): Likewise.
21579 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
21580 arm_arch5e with arm_arch5t and arm_arch5te.
21581 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
21582 (arm_arch5t, arm_arch5te): Declare.
21583 * config/arm/arm-tables.opt: Regenerate.
21584 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
21585 * config/arm/t-multilib: Likewise.
21586 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
21587 instead of arm_arch5.
21588 (*call_reg_thumb1): Likewise.
21589 (*call_value_reg_thumb1_v5): Likewise.
21590 (*call_value_reg_thumb1): Likewise.
21591 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
21593 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
21595 2018-05-18 Martin Liska <mliska@suse.cz>
21597 PR gcov-profile/84846
21598 * doc/gcov.texi: Document -t option of gcov tool.
21600 2018-05-18 Martin Liska <mliska@suse.cz>
21602 PR gcov-profile/84846
21603 * gcov.c (print_usage): Add new -t option.
21604 (process_args): Handle the option.
21605 (generate_results): Use stdout as output when requested by
21608 2018-05-18 Martin Liska <mliska@suse.cz>
21610 PR gcov-profile/84846
21611 * coverage.c (coverage_init): Write PWD to .gcno file.
21612 * doc/gcov.texi: Document how working directory is printed.
21613 * gcov-dump.c (dump_gcov_file): Print PWD.
21614 * gcov.c (output_intermediate_file): Likewise.
21615 (read_graph_file): Read PWD string.
21616 (output_lines): Print PWD.
21618 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21620 PR middle-end/85817
21621 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
21622 for retval and return false if all args to phi are zero.
21624 2018-05-18 Richard Biener <rguenther@suse.de>
21626 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
21628 (evrp_dom_walker::before_dom_children): Call it.
21630 2018-05-18 Richard Biener <rguenther@suse.de>
21632 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
21633 results when processing array refs with variable index.
21635 2018-05-18 Toon Moene <toon@moene.org>
21637 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
21638 directly after that of -floop-interchange. Indicate that both
21639 options are enabled by default when specifying -O3.
21641 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21643 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
21644 iterator. Delete separate integer-mode vec_set<mode> expander.
21645 (aarch64_simd_vec_setv2di): Delete.
21646 (vec_setv2di): Delete.
21647 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
21648 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
21649 the "w, r" alternative.
21651 2018-05-18 Martin Liska <mliska@suse.cz>
21653 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
21654 * tree-pass.h (make_pass_lower_switch_O0): New function.
21655 * tree-switch-conversion.c (node_has_low_bound): Remove.
21656 (node_has_high_bound): Likewise.
21657 (node_is_bounded): Likewise.
21658 (class pass_lower_switch): Make it a template type and create
21660 (pass_lower_switch::execute): Add template argument.
21661 (make_pass_lower_switch): New function.
21662 (make_pass_lower_switch_O0): New function.
21663 (do_jump_if_equal): Remove.
21664 (emit_case_nodes): Simplify to just handle all 3 cases and leave
21665 all the hard work to tree optimization passes.
21667 2018-05-18 Martin Liska <mliska@suse.cz>
21669 * dbgcnt.c (limit_low): Renamed from limit.
21670 (limit_high): New variable.
21671 (dbg_cnt_is_enabled): Check for upper limit.
21672 (dbg_cnt): Adjust dumping.
21673 (dbg_cnt_set_limit_by_index): Add new argument for high
21675 (dbg_cnt_set_limit_by_name): Likewise.
21676 (dbg_cnt_process_single_pair): Parse new format.
21677 (dbg_cnt_process_opt): Use strtok.
21678 (dbg_cnt_list_all_counters): Remove 'value' and add
21680 * doc/invoke.texi: Document changes.
21682 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
21684 * doc/sourcebuild.texi (scalar_all_fma): Document.
21685 * tree.def (FMA_EXPR): Delete.
21686 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
21687 * internal-fn.c (ternary_direct): New macro.
21688 (expand_ternary_optab_fn): Likewise.
21689 (direct_ternary_optab_supported_p): Likewise.
21690 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
21691 * builtins.c (fold_builtin_fma): Delete.
21692 (fold_builtin_3): Don't call it.
21693 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
21694 * expr.c (expand_expr_real_2): Likewise.
21695 * fold-const.c (operand_equal_p): Likewise.
21696 (fold_ternary_loc): Likewise.
21697 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
21698 * gimple.c (DEFTREECODE): Likewise.
21699 * gimplify.c (gimplify_expr): Likewise.
21700 * optabs-tree.c (optab_for_tree_code): Likewise.
21701 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
21702 * tree-eh.c (operation_could_trap_p): Likewise.
21703 (stmt_could_throw_1_p): Likewise.
21704 * tree-inline.c (estimate_operator_cost): Likewise.
21705 * tree-pretty-print.c (dump_generic_node): Likewise.
21706 (op_code_prio): Likewise.
21707 * tree-ssa-loop-im.c (stmt_cost): Likewise.
21708 * tree-ssa-operands.c (get_expr_operands): Likewise.
21709 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
21710 * fold-const-call.h (fold_fma): Delete.
21711 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
21712 CFN_FNMA and CFN_FNMS.
21713 (fold_fma): Delete.
21714 * genmatch.c (combined_fn): New enum.
21715 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
21716 (commutative_op): New function.
21717 (commutate): Use it. Handle more than 2 operands.
21718 (dt_operand::gen_gimple_expr): Use commutative_op.
21719 (parser::parse_expr): Allow :c to be used with non-binary
21720 operators if the commutative operand is known.
21721 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
21722 CFN_FMS, CFN_FNMA and CFN_FNMS.
21723 (backprop::process_assign_use): Remove FMA_EXPR handling.
21724 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
21725 (gen_hsa_fma): New function.
21726 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
21727 IFN_FNMA and IFN_FNMS.
21728 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
21729 * gimple-fold.h (follow_all_ssa_edges): Declare.
21730 * gimple-fold.c (follow_all_ssa_edges): New function.
21731 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
21732 gimple_build interface and use follow_all_ssa_edges to fold the result.
21733 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
21734 instead of checking for optabs directly.
21735 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
21736 rather than FMA_EXPRs.
21737 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
21738 call to IFN_FMA instead of an FMA_EXPR.
21740 2018-05-17 Jim Wilson <jimw@sifive.com>
21742 * expr.c (do_tablejump): When converting index to Pmode, if we have a
21743 sign extended promoted subreg, and the range does not have the sign bit
21744 set, then do a sign extend.
21746 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
21747 test, check for sign extended subreg and/or constant operands, and
21748 do a sign extend in that case.
21750 2018-05-17 Steve Ellcey <sellcey@cavium.com>
21752 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
21753 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
21755 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
21756 Change logics_shift_reg to logics_shift_imm.
21757 (thunderx2t99_fp_loadpair_basic): Delete.
21758 (thunderx2t99_fp_storepair_basic): Delete.
21759 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
21760 (thunderx2t99_asimd_polynomial): Delete.
21761 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
21762 and neon_fp_mul_d_scalar_q.
21763 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
21764 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
21765 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
21766 (thunderx2t99_asimd_lut): Add missing tbl types.
21767 (thunderx2t99_asimd_ext): Delete.
21768 (thunderx2t99_asimd_load1_1_mult): Delete.
21769 (thunderx2t99_asimd_load1_2_mult): Delete.
21770 (thunderx2t99_asimd_load1_ldp): New.
21771 (thunderx2t99_asimd_load1): New.
21772 (thunderx2t99_asimd_load2): Add missing *load2* types.
21773 (thunderx2t99_asimd_load3): New.
21774 (thunderx2t99_asimd_load4): New.
21775 (thunderx2t99_asimd_store1_1_mult): Delete.
21776 (thunderx2t99_asimd_store1_2_mult): Delete.
21777 (thunderx2t99_asimd_store2_mult): Delete.
21778 (thunderx2t99_asimd_store2_onelane): Delete.
21779 (thunderx2t99_asimd_store_stp): New.
21780 (thunderx2t99_asimd_store1): New.
21781 (thunderx2t99_asimd_store2): New.
21782 (thunderx2t99_asimd_store3): New.
21783 (thunderx2t99_asimd_store4): New.
21785 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
21787 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
21788 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
21790 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
21791 Segher Boessenkool <segher@kernel.crashing.org>
21794 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
21797 2018-05-17 Richard Biener <rguenther@suse.de>
21799 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
21800 for pruning loop and prune defs feeding only already visited PHIs.
21802 2018-05-17 Richard Biener <rguenther@suse.de>
21804 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
21806 2018-05-17 Bin Cheng <bin.cheng@arm.com>
21807 Richard Biener <rguenther@suse.de>
21809 PR tree-optimization/85793
21810 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
21811 for VMAT_ELEMENTWISE.
21813 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
21815 * internal-fn.h (lookup_internal_fn): Declare
21816 * internal-fn.c (lookup_internal_fn): New function.
21817 * gimple.c (gimple_build_call_from_tree): Handle calls to
21818 internal functions.
21819 * gimple-pretty-print.c (dump_gimple_call): Print "." before
21820 internal function names.
21821 * tree-pretty-print.c (dump_generic_node): Likewise.
21822 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
21824 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
21826 * gimple-fold.h (gimple_build): Make the function forms take
21827 combined_fn rather than built_in_function.
21828 (gimple_simplify): Likewise.
21829 * gimple-match-head.c (gimple_simplify): Likewise.
21830 * gimple-fold.c (gimple_build): Likewise.
21831 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
21832 rather than gimple_build_call_internal.
21833 (get_initial_defs_for_reduction): Likewise.
21834 (vect_create_epilog_for_reduction): Likewise.
21835 (vectorizable_live_operation): Likewise.
21837 2018-05-17 Martin Liska <mliska@suse.cz>
21839 * gimple-ssa-sprintf.c (format_directive): Do not use
21840 space in between 'G_' and '('.
21842 2018-05-17 Jakub Jelinek <jakub@redhat.com>
21845 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
21846 even if the mask is not all ones.
21849 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
21851 (ix86_gimple_fold_builtin): Likewise.
21854 * config/i386/i386.c: Include tree-vector-builder.h.
21855 (ix86_vector_shift_count): New function.
21856 (ix86_fold_builtin): Fold shift builtins by scalar count.
21857 (ix86_gimple_fold_builtin): Likewise.
21859 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
21860 _mm512_setzero): New intrinsics.
21862 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
21863 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21865 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
21866 code generation for cases where splatting a value is not useful.
21867 * simplify-rtx.c (simplify_ternary_operation): Simplify
21868 vec_merge across a vec_duplicate and a paradoxical subreg forming
21869 a vector mode to a vec_concat.
21871 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
21873 * config.gcc: Support "goldmont-plus".
21874 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
21876 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
21877 PROCESSOR_GOLDMONT_PLUS.
21878 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
21879 (processor_target_table): Add "goldmont-plus".
21880 (PTA_GOLDMONT_PLUS): Define.
21881 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
21882 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
21883 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
21884 (fold_builtin_cpu): Add "goldmont-plus".
21885 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
21886 (ix86_option_override_internal): Add "goldmont-plus".
21887 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
21888 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
21889 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
21890 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
21892 2018-05-17 Richard Biener <rguenther@suse.de>
21894 PR tree-optimization/85757
21895 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
21896 remove defs that only feed that PHI from further processing.
21898 2018-05-16 Jim Wilson <jimw@sifive.com>
21900 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
21902 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
21904 2018-05-16 Mark Wielaard <mark@klomp.org>
21906 * dwarf2out.c (count_index_strings): New function.
21907 (output_indirect_strings): Call count_index_strings and generate
21908 header for dwarf_version >= 5.
21910 2018-05-16 Mark Wielaard <mark@klomp.org>
21912 * dwarf2out.c (dwarf_FORM): New function.
21913 (set_indirect_string): Use dwarf_FORM.
21914 (reset_indirect_string): Likewise.
21915 (size_of_die): Likewise.
21916 (value_format): Likewise.
21917 (output_die): Likewise.
21918 (add_skeleton_AT_string): Likewise.
21919 (output_macinfo_op): Likewise.
21920 (index_string): Likewise.
21921 (output_index_string_offset): Likewise.
21922 (output_index_string): Likewise.
21923 (count_index_strings): Likewise.
21925 2018-05-16 Carl Love <cel@us.ibm.com>
21927 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
21928 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
21930 2018-05-16 Martin Jambor <mjambor@suse.cz>
21932 * ipa-prop.c (ipa_free_all_edge_args): Remove.
21933 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
21935 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
21937 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
21938 (fnma<mode>4): Likewise.
21939 (fms<mode>4): Likewise.
21940 (fnms<mode>4): Likewise.
21941 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
21942 (aarch64_fnma<mode>4): Likewise.
21943 (aarch64_fms<mode>4): Likewise.
21944 (aarch64_fnms<mode>4): Likewise.
21945 (aarch64_fnmadd<mode>4): Likewise.
21947 2018-05-16 Jason Merrill <jason@redhat.com>
21949 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
21951 2018-05-16 Richard Biener <rguenther@suse.de>
21953 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
21954 (dump_stmt_cost): Declare.
21955 (add_stmt_cost): Dump cost we add.
21956 (add_stmt_costs): New function.
21957 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
21958 No longer exported.
21959 (vect_analyze_stmt): Adjust prototype.
21960 (vectorizable_condition): Likewise.
21961 (vectorizable_live_operation): Likewise.
21962 (vectorizable_reduction): Likewise.
21963 (vectorizable_induction): Likewise.
21964 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
21965 cost vector to pass to vectorizable_ and record afterwards.
21966 (vect_model_reduction_cost): Take cost vector argument and adjust.
21967 (vect_model_induction_cost): Likewise.
21968 (vectorizable_reduction): Likewise.
21969 (vectorizable_induction): Likewise.
21970 (vectorizable_live_operation): Likewise.
21971 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
21972 SLP_TREE_NUMBER_OF_VEC_STMTS.
21973 (vect_analyze_slp_cost_1): Remove.
21974 (vect_analyze_slp_cost): Likewise.
21975 (vect_slp_analyze_node_operations): Take visited args and
21976 a target cost vector. Avoid processing already visited stmt sets.
21977 (vect_slp_analyze_operations): Use a local cost vector to gather
21978 costs and register those of non-discarded instances.
21979 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
21980 (vect_schedule_slp_instance): Remove copying of
21981 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
21983 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
21984 adding cost. Record cost entry location.
21985 (vect_prologue_cost_for_slp_op): Function to compute cost of
21986 a constant or invariant generated for SLP vect in the prologue,
21987 split out from vect_analyze_slp_cost_1.
21988 (vect_model_simple_cost): Make static. Adjust for SLP costing.
21989 (vect_model_promotion_demotion_cost): Likewise.
21990 (vect_model_store_cost): Likewise, make static.
21991 (vect_model_load_cost): Likewise.
21992 (vectorizable_bswap): Add cost vector arg and adjust.
21993 (vectorizable_call): Likewise.
21994 (vectorizable_simd_clone_call): Likewise.
21995 (vectorizable_conversion): Likewise.
21996 (vectorizable_assignment): Likewise.
21997 (vectorizable_shift): Likewise.
21998 (vectorizable_operation): Likewise.
21999 (vectorizable_store): Likewise.
22000 (vectorizable_load): Likewise.
22001 (vectorizable_condition): Likewise.
22002 (vectorizable_comparison): Likewise.
22003 (can_vectorize_live_stmts): Likewise.
22004 (vect_analyze_stmt): Likewise.
22005 (vect_transform_stmt): Adjust calls to vectorizable_*.
22006 * tree-vectorizer.c: Include gimple-pretty-print.h.
22007 (dump_stmt_cost): New function.
22009 2018-05-16 Richard Biener <rguenther@suse.de>
22011 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
22012 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
22013 * tree-ssa-dse.c: Include tree-ssa-loop.h.
22014 (check_name): New callback.
22015 (dse_classify_store): Track cycles via a visited bitmap of PHI
22016 defs and simplify handling of in-loop and across loop dead stores
22017 and properly fail for loop-variant refs. Handle byte-tracking with
22018 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
22021 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
22023 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
22024 (vect_get_mask_type_for_stmt): Likewise.
22025 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
22027 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
22028 to determine the statement's vector type and the vector type that
22029 should be used for calculating nunits. Deal with cases in which
22030 the type has to be deferred.
22031 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
22032 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
22033 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
22034 (vect_determine_vf_for_stmt): New functions, split out from...
22035 (vect_determine_vectorization_factor): ...here.
22036 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
22037 (vect_get_mask_type_for_stmt): New functions, split out from
22038 vect_determine_vectorization_factor.
22040 2018-05-16 Richard Biener <rguenther@suse.de>
22042 * tree-cfg.c (verify_gimple_assign_ternary): Properly
22043 verify the [VEC_]COND_EXPR embedded comparison.
22045 2018-05-15 Martin Sebor <msebor@redhat.com>
22047 PR tree-optimization/85753
22048 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
22049 RECORD_TYPE in addition to ARRAY_TYPE.
22051 2018-05-15 Martin Sebor <msebor@redhat.com>
22053 PR middle-end/85643
22054 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
22056 2018-05-15 Richard Biener <rguenther@suse.de>
22058 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
22059 add by_clobber_p one. Change algorithm to collect all defs
22060 representing uses we need to walk and try reducing them to
22061 a single one before failing.
22062 (dse_dom_walker::dse_optimize_stmt): Adjust.
22064 2018-05-13 Mark Wielaard <mark@klomp.org>
22066 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
22067 (size_of_loc_descr): Likewise.
22068 (output_loc_operands): Likewise.
22069 (output_loc_operands_raw): Likewise.
22070 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
22071 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
22072 (hash_loc_operands): Likewise.
22073 (compare_loc_operands): Likewise.
22075 2018-05-14 Mark Wielaard <mark@klomp.org>
22077 * dwarf2out.c (count_index_addrs): New function.
22078 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
22080 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22082 PR tree-optimization/83648
22083 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
22084 return value as malloc candidate.
22086 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22089 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
22090 param as true in call to suggest_attribute.
22092 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
22094 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
22095 -mreadonly-in-sdata.
22097 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22099 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
22101 (aarch64_crypto_aesd_fused): Likewise.
22103 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
22105 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
22106 (movsi_aarch64): Likewise.
22107 (load_pairsi): Likewise.
22108 (load_pairdi): Likewise.
22109 (store_pairsi): Likewise.
22110 (store_pairdi): Likewise.
22111 (load_pairsf): Likewise.
22112 (load_pairdf): Likewise.
22113 (store_pairsf): Likewise.
22114 (store_pairdf): Likewise.
22115 (zero_extend): Likewise.
22116 (trunc): Swap alternatives.
22117 (fcvt_target): Add '?' to prefer w over r.
22119 2018-05-14 Jakub Jelinek <jakub@redhat.com>
22122 * config/i386/i386.md: Disallow non-commutative arithmetics in
22123 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
22124 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
22125 in the peephole2 before it.
22127 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
22129 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
22130 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
22131 (ix86_handle_option): Handle -mcldemote.
22132 * config.gcc: New header.
22133 * config/i386/cldemoteintrin.h: New file.
22134 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
22135 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
22137 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
22138 OPTION_MASK_ISA_CLDEMOTE.
22139 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
22140 (ix86_valid_target_attribute_inner_p): Ditto.
22141 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
22142 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
22143 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
22144 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
22145 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
22147 * config/i386/i386.opt: Add -mcldemote.
22148 * config/i386/x86intrin.h: New header.
22149 * doc/invoke.texi: Add -mcldemote.
22151 2018-05-14 Richard Biener <rguenther@suse.de>
22153 * doc/match-and-simplify.texi: Adjust :s documentation.
22155 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
22157 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
22158 intended memcpy size.
22159 (REORDER_45): Likewise.
22161 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
22163 * sort.cc: New file.
22164 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
22165 * vec.c (qsort_chk): Use gcc_qsort.
22166 * Makefile.in (OBJS-libcommon): Add sort.o.
22167 (build/sort.o): New target. Use it...
22168 (BUILD_RTL): ... here, and...
22169 (build/gencfn-macros): ... here, and...
22170 (build/genmatch): ... here.
22172 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
22173 Chung-Ju Wu <jasonwucj@gmail.com>
22175 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
22176 * config/nds32/nds32-graywolf.md: New file.
22177 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
22178 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
22180 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
22181 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
22182 * config/nds32/nds32.md (pipeline_model): Add graywolf.
22183 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
22184 * config/nds32/pipelines.md: Include n15 settings.
22186 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
22187 Chung-Ju Wu <jasonwucj@gmail.com>
22189 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
22190 * config/nds32/nds32-n13.md: New file.
22191 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
22192 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
22194 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
22195 * config/nds32/nds32.md (pipeline_model): Add n13.
22196 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
22197 * config/nds32/pipelines.md: Include n13 settings.
22199 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
22200 Chung-Ju Wu <jasonwucj@gmail.com>
22202 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
22203 * config/nds32/nds32-n10.md: New file.
22204 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
22205 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
22207 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
22208 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
22209 * config/nds32/nds32.md (pipeline_model): Add n10.
22210 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
22211 * config/nds32/pipelines.md: Include n10 settings.
22213 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
22214 Kito Cheng <kito.cheng@gmail.com>
22215 Chung-Ju Wu <jasonwucj@gmail.com>
22217 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
22218 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
22219 Add enum values for DSP extension instructions.
22220 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
22222 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
22223 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
22224 New code iterators.
22225 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
22226 * config/nds32/nds32-dspext.md: New file for DSP implementation.
22227 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
22228 * config/nds32/nds32-intrinsic.md: Likewise.
22229 * config/nds32/nds32_intrinsic.h: Likewise.
22230 * config/nds32/nds32-md-auxiliary.c: Likewise.
22231 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
22232 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
22233 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
22234 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
22235 * config/nds32/nds32-protos.h: New declarations for DSP extension.
22236 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
22237 TYPE_DMAC in switch statement.
22238 * config/nds32/nds32.c: New checking and implementation for DSP
22239 extension instructions.
22240 * config/nds32/nds32.h: Likewise.
22241 * config/nds32/nds32.md: Likewise.
22242 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
22243 * config/nds32/predicates.md: Implement new predicates for DSP
22246 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22248 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
22249 Reformat alternatives and attributes so it is easier to identify
22250 which constraints/attributes go with which instruction.
22251 (mov<mode>_hardfloat32, FMOVE64): Likewise.
22252 (mov<mode>_softfloat32, FMOVE64): Likewise.
22253 (mov<mode>_hardfloat64, FMOVE64): Likewise.
22254 (mov<mode>_softfloat64, FMOVE64): Likewise.
22256 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
22258 * doc/extend.texi (PowerPC Built-in Functions): Rename this
22260 (Basic PowerPC Built-in Functions): The new name of the
22261 subsection previously known as "PowerPC Built-in Functions".
22262 (Basic PowerPC Built-in Functions Available on all Configurations):
22264 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
22265 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
22266 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
22267 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
22269 2018-05-11 Martin Jambor <mjambor@suse.cz>
22272 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
22275 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
22278 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
22280 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
22282 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
22283 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
22284 (ix86_handle_option): Handle -mwaitpkg.
22285 * config.gcc: New header.
22286 * config/i386/cpuid.h (bit_WAITPKG): New bit.
22287 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
22288 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
22290 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
22291 OPTION_MASK_ISA_WAITPKG.
22292 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
22293 (ix86_option_override_internal): Add PTA_WAITPKG.
22294 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
22295 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
22296 IX86_BUILTIN_TPAUSE.
22297 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
22298 __builtin_ia32_umwait and __builtin_ia32_tpause.
22299 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
22300 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
22301 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
22302 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
22303 UNSPECV_TPAUSE): New.
22304 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
22305 * config/i386/i386.opt: Add -mwaitpkg.
22306 * config/i386/waitpkgintrin.h: New file.
22307 * config/i386/x86intrin.h: New header.
22308 * doc/invoke.texi: Add -mwaitpkg.
22310 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
22313 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
22315 (cortex-m0): Use armv6s-m isa.
22316 (cortex-m0plus): Likewise.
22317 (cortex-m1): Likewise.
22318 (cortex-m0.small-multiply): Likewise.
22319 (cortex-m0plus.small-multiply): Likewise.
22320 (cortex-m1.small-multiply): Likewise.
22322 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
22323 Jakub Jelinek <jakub@redhat.com>
22325 PR tree-optimization/85692
22326 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
22327 source permute as well.
22329 2018-05-11 Martin Liska <mliska@suse.cz>
22332 * doc/extend.texi: Document LLVM style format for no_sanitize
22335 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
22337 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
22338 mode_supports_vsx_dform_quad to mode_supports_dq_form.
22339 (mode_supports_vsx_dform_quad): Likewise.
22340 (mode_supports_vmx_dform): Move these functions to be next to the
22341 other mode_supports functions.
22342 (mode_supports_dq_form): Likewise.
22343 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
22344 mode_supports_dq_form.
22345 (reg_offset_addressing_ok_p): Likewise.
22346 (offsettable_ok_by_alignment): Likewise.
22347 (rs6000_legitimate_offset_address_p): Likewise.
22348 (legitimate_lo_sum_address_p): Likewise.
22349 (rs6000_legitimize_address): Likewise.
22350 (rs6000_legitimize_reload_address): Likewise.
22351 (rs6000_secondary_reload_inner): Likewise.
22352 (rs6000_preferred_reload_class): Likewise.
22353 (rs6000_output_move_128bit): Likewise.
22355 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
22357 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
22358 Generate SImode target register for null target.
22359 <case IX86_BUILTIN_XGETBV>: Ditto.
22360 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
22361 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
22363 2018-05-10 Carl Love <cel@us.ibm.com>
22365 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
22366 dcbtt and dcbtstt if operands[2] is 0.
22368 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
22371 * config/i386/sse.md (usadv64qi): New expander.
22373 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
22375 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
22376 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
22377 -maltivec=be support.
22378 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
22379 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
22380 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22381 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
22382 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
22383 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
22384 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
22385 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
22386 altivec_vsumsws): Adjust.
22387 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
22388 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
22389 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
22391 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
22392 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
22393 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
22394 (altivec_lve<VI_char>x): Delete expand.
22395 (*altivec_lve<VI_char>x_internal): Rename to...
22396 (altivec_lve<VI_char>x): ... this.
22397 (altivec_lvxl_<mode>): Delete expand.
22398 (*altivec_lvxl_<mode>_internal): Rename to ...
22399 (altivec_lvxl_<mode>): ... this.
22400 (altivec_stvxl_<mode>): Delete expand.
22401 (*altivec_stvxl_<mode>_internal): Rename to ...
22402 (altivec_stvxl_<mode>): ... this.
22403 (altivec_stve<VI_char>x): Delete expand.
22404 (*altivec_stve<VI_char>x_internal): Rename to ...
22405 (altivec_stve<VI_char>x): ... this.
22406 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
22407 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
22408 reduc_plus_scal_<mode>): Adjust.
22409 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
22411 (rs6000_cpu_cpp_builtins): Adjust.
22412 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
22413 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
22414 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
22415 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
22416 -maltivec=be support.
22417 (rs6000_split_vec_extract_var): Adjust.
22418 (rs6000_split_v4si_init): Adjust.
22419 (swap_selector_for_mode): Delete.
22420 (altivec_expand_lvx_be, altivec_expand_stvx_be,
22421 altivec_expand_stvex_be): Delete.
22422 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
22423 -maltivec=be support.
22424 (rs6000_gimple_fold_builtin): Ditto.
22425 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
22427 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
22428 (TARGET_DIRECT_MOVE_64BIT): Adjust.
22429 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
22430 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
22431 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
22432 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
22433 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
22434 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
22435 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
22436 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
22437 anonymous split): Adjust.
22438 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
22439 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
22441 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
22443 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
22444 when --with-gxx-include-dir is also specified.
22445 * configure: Regenerate.
22447 2018-05-09 Jim Wilson <jimw@sifive.com>
22450 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
22451 * config/riscv/t-withmultilib: New.
22452 * config/riscv/withmultilib.h: New.
22453 * doc/install.texi: Document RISC-V --with-multilib-list support.
22455 2018-05-09 Richard Biener <rguenther@suse.de>
22457 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
22459 (vect_bb_vectorization_profitable_p): Adjust. Compute
22460 actual scalar cost using the cost vector and the add_stmt_cost
22463 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
22465 PR rtl-optimization/85645
22466 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
22467 in the REG_CFA_REGISTER note for LR, don't leave it empty.
22469 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
22471 PR rtl-optimization/85645
22472 * shrink-wrap.c (spread_components): Return a boolean saying if
22473 anything was changed.
22474 (try_shrink_wrapping_separate): Iterate spread_components until
22475 nothing changes anymore.
22477 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
22479 PR rtl-optimization/85645
22480 * regrename.c (build_def_use): Also kill the chains that include the
22481 destination of a REG_CFA_REGISTER note.
22483 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
22485 PR rtl-optimization/85645
22486 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
22487 insn that has a REG_CFA_REGISTER note.
22489 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
22491 * cfgexpand.c (expand_clobber): New function.
22492 (expand_gimple_stmt_1): Use it.
22493 * tree-vect-stmts.c (vect_clobber_variable): New function,
22495 (vectorizable_simd_clone_call): ...here.
22496 (vectorizable_store): Emit a clobber either side of an
22497 IFN_STORE_LANES sequence.
22498 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
22500 2018-05-09 Tom de Vries <tom@codesourcery.com>
22503 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
22504 (define_insn "trap_if_false"): Add exit after trap.
22506 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
22508 PR rtl-optimization/85638
22509 * bb-reorder.c: Include common/common-target.h.
22510 (create_forwarder_block): New function extracted from...
22511 (fix_up_crossing_landing_pad): ...here. Rename into...
22512 (dw2_fix_up_crossing_landing_pad): ...this.
22513 (sjlj_fix_up_crossing_landing_pad): New function.
22514 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
22515 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
22516 from both partitions and exit the loop after one iteration.
22518 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
22521 * doc/extend.texi (PowerPC Built-in Functions): Rename this
22523 (Basic PowerPC Built-in Functions): The new name of the
22524 subsection previously known as "PowerPC Built-in Functions".
22525 (Basic PowerPC Built-in Functions Available on all Configurations):
22527 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
22529 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
22531 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
22533 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
22536 2018-05-08 Jim Wilson <jimw@sifive.com>
22538 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
22539 (LD_EMUL_SUFFIX): New.
22540 (LINK_SPEC): Use it.
22542 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
22544 * doc/extend.texi (PowerPC Built-in Functions): Rename this
22546 (Basic PowerPC Built-in Functions): The new name of the
22547 subsection previously known as "PowerPC Built-in Functions".
22548 (Basic PowerPC Built-in Functions Available on all Configurations):
22550 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
22552 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
22554 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
22556 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
22559 2018-05-08 Jakub Jelinek <jakub@redhat.com>
22562 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
22563 after cmpelim optimization.
22565 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
22567 * config.gcc: Support "goldmont".
22568 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
22569 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
22570 PROCESSOR_GOLDMONT.
22571 * config/i386/i386.c (m_GOLDMONT): Define.
22572 (processor_target_table): Add "goldmont".
22573 (PTA_GOLDMONT): Define.
22574 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
22575 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
22576 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
22577 (fold_builtin_cpu): Add "goldmont".
22578 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
22579 (ix86_option_override_internal): Add "goldmont".
22580 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
22581 (processor_type): Add PROCESSOR_GOLDMONT.
22582 * config/i386/i386.md: Add CPU "glm".
22583 * config/i386/glm.md: New file.
22584 * config/i386/x86-tune.def: Add m_GOLDMONT.
22585 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
22587 2018-05-08 Jakub Jelinek <jakub@redhat.com>
22590 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
22592 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
22593 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
22594 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
22597 * config/i386/i386.c (ix86_fold_builtin): Handle
22598 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
22601 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
22602 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
22604 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
22607 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
22608 (check_arch): Likewise.
22609 (check_fpu): Return the result rather than printing it.
22610 (end arch): Fix operator precedence.
22611 (end cpu): Likewise.
22612 (END): Print the result from check_fpu.
22614 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
22615 Alan Hayward <alan.hayward@arm.com>
22616 David Sherwood <david.sherwood@arm.com>
22618 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
22619 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
22620 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
22621 (*fcmuo<mode>_and): New patterns.
22623 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
22625 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
22626 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
22627 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
22628 (cmp_op, sve_imm_con): New code attributes.
22629 (SVE_COND_INT_CMP, imm_con): Delete.
22630 (cmp_op): Remove above unspecs from int attribute.
22631 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
22633 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
22634 comparison-specific unspecs.
22635 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
22636 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
22637 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
22638 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
22639 (*vec_fcm<cmp_op><mode>): Rename to...
22640 (*fcm<cmp_op><mode>): ...this and adjust likewise.
22641 (*vec_fcmuo<mode>): Rename to...
22642 (*fcmuo<mode>): ...this and adjust likewise.
22643 (*pred_fcm<cmp_op><mode>): New pattern.
22644 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
22645 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
22647 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
22649 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
22650 (aarch64_emit_sve_predicated_cond): New function.
22651 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
22652 (aarch64_emit_unspec_cond_or): Replace with...
22653 (aarch64_emit_sve_or_conds): ...this new function. Use
22654 aarch64_emit_sve_ptrue_op for the individual comparisons and
22655 aarch64_emit_binop to OR them together.
22656 (aarch64_emit_inverted_unspec_cond): Replace with...
22657 (aarch64_emit_sve_inverted_cond): ...this new function. Use
22658 aarch64_emit_sve_ptrue_op for the comparison and
22659 aarch64_emit_unop to invert the result.
22660 (aarch64_expand_sve_vec_cmp_float): Update after the above
22661 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
22663 2018-05-07 Nathan Sidwell <nathan@acm.org>
22665 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
22666 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
22667 (Backwards Compatibility): Likewise.
22669 2018-05-07 Luis Machado <luis.machado@linaro.org>
22673 2018-05-07 Luis Machado <luis.machado@linaro.org>
22675 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
22676 <prefetch_dynamic_strides>: New const bool field.
22677 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
22678 prefetch_dynamic_strides.
22679 (exynosm1_prefetch_tune): Likewise.
22680 (thunderxt88_prefetch_tune): Likewise.
22681 (thunderx_prefetch_tune): Likewise.
22682 (thunderx2t99_prefetch_tune): Likewise.
22683 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
22685 (aarch64_override_options_internal): Update to set
22686 PARAM_PREFETCH_DYNAMIC_STRIDES.
22687 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
22688 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
22689 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
22690 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
22691 prefetch-dynamic-strides setting.
22693 2018-05-07 Luis Machado <luis.machado@linaro.org>
22695 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
22696 <minimum_stride>: New const int field.
22697 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
22698 minimum_stride field.
22699 (exynosm1_prefetch_tune): Likewise.
22700 (thunderxt88_prefetch_tune): Likewise.
22701 (thunderx_prefetch_tune): Likewise.
22702 (thunderx2t99_prefetch_tune): Likewise.
22703 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
22704 (aarch64_override_options_internal): Update to set
22705 PARAM_PREFETCH_MINIMUM_STRIDE.
22706 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
22707 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
22708 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
22709 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
22710 stride is constant and is below the minimum stride threshold.
22712 2018-05-07 Luis Machado <luis.machado@linaro.org>
22714 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
22717 2018-05-07 Luis Machado <luis.machado@linaro.org>
22719 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
22720 <prefetch_dynamic_strides>: New const bool field.
22721 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
22722 prefetch_dynamic_strides.
22723 (exynosm1_prefetch_tune): Likewise.
22724 (thunderxt88_prefetch_tune): Likewise.
22725 (thunderx_prefetch_tune): Likewise.
22726 (thunderx2t99_prefetch_tune): Likewise.
22727 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
22729 (aarch64_override_options_internal): Update to set
22730 PARAM_PREFETCH_DYNAMIC_STRIDES.
22731 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
22732 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
22733 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
22734 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
22735 prefetch-dynamic-strides setting.
22737 2018-05-07 Luis Machado <luis.machado@linaro.org>
22739 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
22740 <minimum_stride>: New const int field.
22741 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
22742 minimum_stride field.
22743 (exynosm1_prefetch_tune): Likewise.
22744 (thunderxt88_prefetch_tune): Likewise.
22745 (thunderx_prefetch_tune): Likewise.
22746 (thunderx2t99_prefetch_tune): Likewise.
22747 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
22748 (aarch64_override_options_internal): Update to set
22749 PARAM_PREFETCH_MINIMUM_STRIDE.
22750 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
22751 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
22752 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
22753 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
22754 stride is constant and is below the minimum stride threshold.
22756 2018-05-06 Jakub Jelinek <jakub@redhat.com>
22759 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
22760 the type is addressable. Don't force op into register if it has
22763 2018-05-05 Roland McGrath <mcgrathr@google.com>
22766 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
22767 any section for which we don't know a specific type it should have,
22768 regardless of name. Previously this was done only for the exact
22769 names ".init_array", ".fini_array", and ".preinit_array".
22770 (default_elf_asm_named_section): Add comment about
22771 relationship with default_section_type_flags and SECTION_NOTYPE.
22772 (get_section): Don't consider it a type conflict if one side has
22773 SECTION_NOTYPE and the other doesn't, as long as neither has the
22774 SECTION_BSS et al used in the default_section_type_flags logic.
22776 2018-05-05 Tom de Vries <tom@codesourcery.com>
22779 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
22780 (workaround_barsyncs): New function.
22781 (nvptx_reorg): Use workaround_barsyncs.
22782 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
22783 (define_expand "nvptx_membar_cta"): New define_expand.
22784 (define_insn "*nvptx_membar_cta"): New insn.
22786 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22788 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
22789 To improve optimization opportunities.
22790 * builtin-types.def: The new needed builtin types for the above.
22792 2018-05-04 Richard Biener <rguenther@suse.de>
22794 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
22795 * gimple-ssa-store-merging.c
22796 (imm_store_chain_info::output_merged_store): Remove redundant create,
22797 release split_store vector contents on failure.
22798 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
22799 scalar stmt vector on cache hit.
22801 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
22803 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22805 * config.gcc (powerpc-xilinx-eabi*): Remove.
22806 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
22808 (fusion_addis_mem_combo_load): Ditto.
22809 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
22811 (rs6000_cpu_cpp_builtins): Ditto.
22812 * config/rs6000/rs6000-linux.c
22813 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
22814 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
22815 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
22817 (rs6000_setup_reg_addr_masks): Ditto.
22818 (rs6000_init_hard_regno_mode_ok): Ditto.
22819 (rs6000_option_override_internal): Ditto.
22820 (legitimate_lo_sum_address_p): Ditto.
22821 (rs6000_legitimize_address): Ditto.
22822 (rs6000_legitimize_reload_address): Ditto.
22823 (rs6000_legitimate_address_p): Ditto.
22824 (abi_v4_pass_in_fpr): Ditto.
22825 (setup_incoming_varargs): Ditto.
22826 (rs6000_gimplify_va_arg): Ditto.
22827 (rs6000_split_multireg_move): Ditto.
22828 (rs6000_savres_strategy): Ditto.
22829 (rs6000_emit_prologue_components): Ditto.
22830 (rs6000_emit_epilogue_components): Ditto.
22831 (rs6000_emit_prologue): Ditto.
22832 (rs6000_emit_epilogue): Ditto.
22833 (rs6000_elf_file_end): Ditto.
22834 (rs6000_function_value): Ditto.
22835 (rs6000_libcall_value): Ditto.
22836 * config/rs6000/rs6000.h: Ditto.
22837 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
22838 (TARGET_MINMAX): ... this. New.
22839 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
22840 * config/rs6000/rs6000.md: Remove Xilinx FP support.
22841 (*movsi_internal1_single): Delete.
22842 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
22843 mfpu=, mxilinx-fpu): Delete.
22844 * config/rs6000/singlefp.h: Delete.
22845 * config/rs6000/sysv4.h: Remove Xilinx FP support.
22846 * config/rs6000/t-rs6000: Ditto.
22847 * config/rs6000/t-xilinx: Delete.
22848 * config/rs6000/titan.md: Adjust for fp_type removal.
22849 * config/rs6000/vsx.md: Remove Xilinx FP support.
22850 (VStype_simple): Delete.
22851 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
22852 * config/rs6000/xfpu.h: Delete.
22853 * config/rs6000/xfpu.md: Delete.
22854 * config/rs6000/xilinx.h: Delete.
22855 * config/rs6000/xilinx.opt: Delete.
22856 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
22857 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
22859 2018-05-04 Tom de Vries <tom@codesourcery.com>
22862 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
22865 2018-05-04 Richard Biener <rguenther@suse.de>
22867 PR middle-end/85627
22868 * tree-complex.c (update_complex_assignment): We are always in SSA form.
22869 (expand_complex_div_wide): Likewise.
22870 (expand_complex_operations_1): Likewise.
22871 (expand_complex_libcall): Preserve EH info of the original stmt.
22872 (tree_lower_complex): Handle removed blocks.
22873 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
22874 on complex multiplication and division libcall builtins.
22876 2018-05-04 Richard Biener <rguenther@suse.de>
22878 PR middle-end/85574
22879 * fold-const.c (negate_expr_p): Restrict negation of operand
22880 zero of a division to when we know that can happen without
22882 (fold_negate_expr_1): Likewise.
22884 2018-05-04 Jakub Jelinek <jakub@redhat.com>
22887 * real.h (real_nextafter): Declare.
22888 * real.c (real_nextafter): New function.
22889 * fold-const-call.c (fold_const_nextafter): New function.
22890 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
22891 CASE_CFN_NEXTTOWARD.
22892 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
22893 even when arg1_mode is different from arg0_mode.
22895 2018-05-03 Nathan Sidwell <nathan@acm.org>
22897 * doc/extend.texi (Deprecated Features): Remove
22898 -ffriend-injection.
22899 (Backwards Compatibility): Likewise.
22900 * doc/invoke.texi (C++ Language Options): Likewise.
22901 (C++ Dialect Options): Likewise.
22903 2018-05-03 Jakub Jelinek <jakub@redhat.com>
22906 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
22907 _mm512_mask_mullox_epi64): New intrinsics.
22909 2018-05-03 Tom de Vries <tom@codesourcery.com>
22912 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
22913 dump files): Add offload-tree.
22915 2018-05-03 Richard Biener <rguenther@suse.de>
22917 PR tree-optimization/85615
22918 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
22919 to loops not nested in BBs loop father to avoid creating multi-entry
22922 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22924 PR tree-optimization/70291
22925 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
22926 arguments. Change return type to tree. Emit libcall as a new
22927 statement rather than replacing existing one when inplace_p is true.
22928 (expand_complex_multiplication_components): New function.
22929 (expand_complex_multiplication): Expand floating-point complex
22930 multiplication using the above.
22931 (expand_complex_division): Rename inner_type parameter to type.
22932 Update expand_complex_libcall call-site.
22933 (expand_complex_operations_1): Update expand_complex_multiplication
22934 and expand_complex_division call-sites.
22936 2018-05-02 Jakub Jelinek <jakub@redhat.com>
22939 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
22940 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
22941 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
22942 the highest significant bit of the shift count mask is clear. In
22943 check whether and[sq]i3 is needed verify that all significant bits
22944 of the shift count other than the highest are set.
22946 2018-05-02 Tom de Vries <tom@codesourcery.com>
22949 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
22950 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
22951 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
22952 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
22953 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
22954 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
22955 __builtin_goacc_parlevel_size.
22957 2018-05-02 Richard Biener <rguenther@suse.de>
22959 PR tree-optimization/85597
22960 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
22961 do not use split vect_get_vec_defs call but call vect_get_slp_defs
22964 2018-05-02 Tom de Vries <tom@codesourcery.com>
22967 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
22968 dump files): Add ltrans-tree.
22970 2018-05-02 Tom de Vries <tom@codesourcery.com>
22973 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
22974 dump files): Add wpa-ipa.
22976 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
22978 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
22979 powerpc*-*-linux*paired* target.
22980 * config/rs6000/750cl.h: Delete.
22981 * config/rs6000/paired.h: Delete.
22982 * config/rs6000/paired.md: Delete.
22983 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
22985 * config/rs6000/rs6000-builtin.def: Remove paired float support.
22986 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
22987 comment. Remove paired float support.
22988 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
22989 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
22991 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
22992 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
22994 * config/rs6000/rs6000.c: Remove paired float support.
22995 (paired_expand_vector_init, paired_expand_vector_move,
22996 paired_emit_vector_compare, paired_emit_vector_cond_expr,
22997 (paired_expand_lv_builtin, paired_expand_stv_builtin,
22998 paired_expand_builtin, paired_expand_predicate_builtin,
22999 paired_init_builtins): Delete.
23000 * config/rs6000/rs6000.h: Remove paired float support.
23001 * config/rs6000/rs6000.md: Remove paired float support.
23002 (move_from_CR_ov_bit): Delete.
23003 * config/rs6000/rs6000.opt (mpaired): Delete.
23004 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
23005 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
23007 2018-05-02 Richard Biener <rguenther@suse.de>
23009 PR middle-end/85567
23010 * gimplify.c (gimplify_save_expr): When in SSA form allow
23011 SAVE_EXPRs to compute to SSA vars.
23013 2018-05-02 Jakub Jelinek <jakub@redhat.com>
23016 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
23017 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
23018 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
23019 clobber operands[2], instead use a new pseudo. Formatting fixes.
23021 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
23023 PR tree-optimization/85586
23024 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
23025 exit early for statements in the same group if the accesses are
23028 2018-05-02 Tom de Vries <tom@codesourcery.com>
23031 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
23034 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
23036 PR tree-optimization/85143
23037 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
23039 2018-05-01 Tom de Vries <tom@codesourcery.com>
23042 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
23043 not found" error message.
23045 2018-05-01 Tom de Vries <tom@codesourcery.com>
23048 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
23049 * vec.c (test_ordered_remove_if): New function.
23050 (vec_c_tests): Call test_ordered_remove_if.
23051 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
23052 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
23053 * tree-vect-patterns.c (vect_pattern_recog_1): Use
23054 VEC_ORDERED_REMOVE_IF.
23056 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23058 PR tree-optimization/82665
23059 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
23060 pointer subtraction where arguments come from a memchr call.
23062 2018-05-01 Jakub Jelinek <jakub@redhat.com>
23064 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
23065 --push-state --as-needed and --pop-state instead of --as-needed and
23066 --no-as-needed if ld supports it.
23067 * configure: Regenerated.
23070 * doc/install.texi2html: Replace _002d with - and _002a with * in
23071 generated html files using sed.
23073 2018-04-30 David Malcolm <dmalcolm@redhat.com>
23076 * gcc-rich-location.c (blank_line_before_p): New function.
23077 (use_new_line): New function.
23078 (gcc_rich_location::add_fixit_insert_formatted): New function.
23079 * gcc-rich-location.h
23080 (gcc_rich_location::add_fixit_insert_formatted): New function.
23082 2018-04-30 David Malcolm <dmalcolm@redhat.com>
23084 * selftest.c (assert_streq): Rename "expected" and "actual" to
23085 "val1" and "val2". Extend NULL-handling to cover both inputs
23086 symmetrically, while still requiring both to be non-NULL for a pass.
23087 * selftest.h (assert_streq): Rename "expected" and "actual" to
23089 (ASSERT_EQ): Likewise.
23090 (ASSERT_EQ_AT): Likewise.
23091 (ASSERT_KNOWN_EQ): Likewise.
23092 (ASSERT_KNOWN_EQ_AT): Likewise.
23093 (ASSERT_NE): Likewise.
23094 (ASSERT_MAYBE_NE): Likewise.
23095 (ASSERT_MAYBE_NE_AT): Likewise.
23096 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
23097 the assertion to pass.
23098 (ASSERT_STREQ_AT): Likewise.
23100 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
23102 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
23103 interaction with -pie.
23105 2018-04-30 David Malcolm <dmalcolm@redhat.com>
23107 * selftest.h: Fix alphabetization of per-source-file selftest
23110 2018-04-30 Jason Merrill <jason@redhat.com>
23112 PR c++/61982 - dead stores to destroyed objects.
23113 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
23116 2018-04-30 Jason Merrill <jason@redhat.com>
23118 * tree.c (build_clobber): New.
23119 * tree.h: Declare it.
23120 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
23122 2018-04-30 David Malcolm <dmalcolm@redhat.com>
23124 * diagnostic-show-locus.c (layout::layout): Update for
23125 location_get_source_line returning a char_span.
23126 (struct char_span): Move to input.h.
23127 (struct correction): Update for fields in char_span becoming
23129 (struct source_line): Update for location_get_source_line
23130 returning a char_span.
23131 (layout::print_line): Likewise.
23132 * edit-context.c (edited_file::print_content): Likewise.
23133 (edited_file::print_diff_hunk): Likewise.
23134 (edited_file::print_run_of_changed_lines): Likewise.
23135 (edited_file::get_num_lines): Likewise.
23136 (edited_line::edited_line): Likewise.
23137 * final.c (asm_show_source): Likewise.
23138 * input.c (location_get_source_line): Convert return type
23139 from const char * to char_span, losing the final "line_len"
23141 (dump_location_info): Update for the above.
23142 (get_substring_ranges_for_loc): Likewise. Use a char_span
23143 when handling the literal within the line.
23144 (test_reading_source_line): Update for location_get_source_line
23145 returning a char_span.
23146 * input.h (class char_span): Move here from
23147 diagnostic-show-locus.c, converting from a struct to a class.
23148 Make data members private.
23149 (char_span::operator bool): New.
23150 (char_span::length): New.
23151 (char_span::get_buffer): New.
23152 (char_span::operator[]): New.
23153 (char_span::subspan): Make const.
23154 (char_span::xstrdup): New.
23155 (location_get_source_line): Convert return type from const char *
23156 to char_span, losing the final "line_size" param.
23158 2018-04-30 Jan Hubicka <jh@suse.cz>
23160 * lto-wrapper.c (ltrans_priorities): New static var.
23161 (cmp_priority): New.
23162 (run_gcc): Read priorities and if doing parallel build order
23163 the Makefile by them.
23165 2018-04-30 David Malcolm <dmalcolm@redhat.com>
23167 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
23169 2018-04-30 Richard Biener <rguenther@suse.de>
23171 * tree-cfg.c (verify_address): Remove base argument, add
23172 flag whether to check TREE_ADDRESSABLE and do that.
23173 (verify_expr): Remove.
23174 (verify_types_in_gimple_reference): Add pieces from verify_expr.
23175 (verify_gimple_assign_single): Likewise.
23176 (verify_gimple_switch): Likewise.
23177 (verify_expr_location_1): Dereference tp once. Add (disabled)
23178 piece from verify_expr.
23179 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
23181 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
23183 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
23185 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
23187 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
23188 (small_data_pattern): Likewise.
23189 (arc_rewrite_small_data): Likewise.
23190 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
23191 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
23192 (get_symbol_alignment): New function.
23193 (legitimate_small_data_address_p): Likewise.
23194 (legitimate_scaled_address): Update, call
23195 legitimate_small_data_address_p.
23196 (output_sdata): New static variable.
23197 (arc_print_operand): Update how we handle small data operands.
23198 (arc_print_operand_address): Likewise.
23199 (arc_legitimate_address_p): Update, use
23200 legitimate_small_data_address_p.
23201 (arc_rewrite_small_data_p): Remove.
23202 (arc_rewrite_small_data_1): Likewise.
23203 (arc_rewrite_small_data): Likewise.
23204 (small_data_pattern): Likewise.
23205 (compact_sda_memory_operand): Update to use
23206 legitimate_small_data_address_p and get_symbol_alignment.
23207 (prepare_move_operands): Don't rewite sdata pattern.
23208 (prepare_extend_operands): Remove.
23209 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
23211 (zero_extendqisi2): Likewise.
23212 (zero_extendhisi2): Likewise.
23213 (extendqihi2): Likewise.
23214 (extendqisi2): Likewise.
23215 (extendhisi2): Likewise.
23216 (addsi3): Likewise.
23217 (subsi3): Likewise.
23218 (andsi3): Likewise.
23219 * config/arc/constraints.md (Usd): Change it to memory constraint.
23221 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
23223 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
23224 as source of std instructions.
23225 * config/arc/arc.md (movsi_insn): Update pattern predicate to
23226 allow 6-bit constants as source for store instructions.
23227 (movdi_insn): Update instruction pattern to allow 6-bit constants
23228 as source for store instructions.
23230 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
23232 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
23234 2018-04-30 Nathan Sidwell <nathan@acm.org>
23235 Sandra Loosemore <sandra@codesourcery.com>
23237 * dumpfile.c (dump_open): Allow '-' for stdout.
23238 * doc/invoke.texi (Developer Options): Document dump filename
23239 determination early. Document stdin/stdout selection.
23241 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
23243 Microblaze Target: PIC data text relative
23245 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
23246 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
23248 * config/microblaze/microblaze.h (microblaze_constant_address_p):
23249 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
23250 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
23251 New addressing mode for data-text relative position indepenedent code.
23252 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
23253 'ADDRESS_SYMBOLIC_TXT_REL'.
23254 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
23255 (microblaze_legitimate_pic_operand): Exclude function calls from
23256 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
23257 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
23259 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
23260 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
23261 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
23262 for 'address + offset'.
23263 (microblaze_expand_prologue): Add new function prologue call for
23265 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
23266 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
23267 table in case of TARGET_PIC_DATA_TEXT_REL.
23268 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
23269 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
23270 Add new macros 'UNSPEC_TEXT',
23271 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
23272 + exclude function calls from 'UNSPEC_PLT' in case of data text
23274 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
23275 new target hook for generating address diff vector tables in case of
23277 * doc/tm.texi : Regenerate.
23278 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
23279 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
23280 of addr diff vector generation.
23281 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
23282 target hook definition.
23283 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
23284 Add default function for generate_pic_addr_diff_vec -> flag_pic.
23285 * doc/invoke.texi (Add new pic option): Add new microblaze pic
23286 option for data text relative.
23288 2018-04-30 Richard Biener <rguenther@suse.de>
23290 * tree-chrec.h (evolution_function_is_constant_p): Remove
23292 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
23294 2018-04-30 Richard Biener <rguenther@suse.de>
23297 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
23299 2018-04-30 Richard Biener <rguenther@suse.de>
23301 PR tree-optimization/28364
23302 PR tree-optimization/85275
23303 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
23304 copying first exit test.
23306 2018-04-28 Mark Wielaard <mark@klomp.org>
23308 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
23309 dwarf_version >= 5.
23310 (dwarf_AT): Handle DW_AT_addr_base.
23311 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
23313 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
23316 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
23317 (*ashl<dwi>3_doubleword_mask_1): Ditto.
23318 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
23319 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
23321 2018-04-28 Richard Biener <rguenther@suse.de>
23323 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
23324 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
23325 to reflect use. Only add interesting stmts.
23327 2018-04-27 Martin Jambor <mjambor@suse.cz>
23330 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
23331 the jump function allows for passing through aggregate values.
23333 2018-04-27 David Malcolm <dmalcolm@redhat.com>
23335 * input.h (in_system_header_at): Convert from macro to inline
23337 (from_macro_expansion_at): Likewise.
23338 (from_macro_definition_at): Likewise.
23340 2018-04-27 Jeff Law <law@redhat.com>
23342 * config.gcc: Mark tile* targets as deprecated/obsolete.
23344 2018-04-27 Richard Biener <rguenther@suse.de>
23346 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
23349 2018-04-27 Richard Biener <rguenther@suse.de>
23351 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
23353 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
23355 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
23356 with Yd constraint. Set "preferred_for_speed" attribute from
23357 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
23358 with Yd constraint.
23359 (*movdi_internal): Ditto.
23360 (movti_interunit splitters): Remove
23361 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
23362 (movdi_interunit splitters): Ditto.
23363 * config/i386/constraints.md (Ye): Remove.
23364 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
23366 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23369 * config/aarch64/constraints.md (Usg): Limit to 31.
23370 (Usj): Limit to 63.
23372 2018-04-27 Jakub Jelinek <jakub@redhat.com>
23374 PR tree-optimization/85529
23375 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
23376 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
23377 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
23378 zero extension or masking of the MSB bit.
23379 (optimize_range_tests): Add FIRST_BB argument, pass it through
23380 to optimize_range_tests_var_bound.
23381 (maybe_optimize_range_tests, reassociate_bb): Adjust
23382 optimize_range_tests callers.
23384 2018-04-26 Richard Biener <rguenther@suse.de>
23385 Jakub Jelinek <jakub@redhat.com>
23387 * cgraph.h (symbol_table): Just declare debug method here.
23388 * symtab.c (symbol_table::debug): Define.
23390 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
23392 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
23394 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
23396 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
23397 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
23398 (*movdi_internal): Substitute Yi and Yj constraint with x
23399 and Ym and Yn constraint with y constraint. Update "isa"
23400 attribute and set "preferred_for_speed" attribute from
23401 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
23402 (*movsi_internal): Ditto.
23403 (*movdf_internal): Ditto.
23404 (*movsf_internal): Ditto.
23405 (*zero_extendsidi2): Ditto.
23406 * config/i386/sse.md (vec_set<mode>_0): Ditto.
23407 (sse2_loadld): Ditto.
23408 (*vec_extract<ssevecmodelower>_0): Ditto.
23409 (*vec_extractv4si_0_zext_sse4): Ditto.
23410 (vec_concatv2di): Ditto.
23411 (*vec_dup<mode>): Ditto.
23412 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
23413 * config/i386/constraints.md (Yi): Remove.
23418 2018-04-26 Nathan Sidwell <nathan@acm.org>
23420 * dumpfile.c (dump_open): New.
23421 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
23422 (dump_finish): Detect stdio/stderr by value not name.
23424 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
23426 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
23428 2018-04-26 Tom de Vries <tom@codesourcery.com>
23431 * config/nvptx/nvptx.c (verify_neutering_jumps)
23432 (verify_neutering_labels): New function
23433 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
23435 2018-04-26 Tom de Vries <tom@codesourcery.com>
23438 * config/nvptx/nvptx.c (needs_neutering_p): New function.
23439 (nvptx_single): Use needs_neutering_p to skip over insns that do not
23442 2018-04-26 Richard Biener <rguenther@suse.de>
23443 Tom de Vries <tom@codesourcery.com>
23446 * lto-streamer-out.c (output_function): Fixup loops if required to match
23447 discovery done in the reader.
23449 2018-04-26 Richard Biener <rguenther@suse.de>
23451 PR tree-optimization/85116
23452 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
23453 have a loop exit from the single latch predecessor. Remove
23454 case of header with just condition.
23455 (ch_base::copy_headers): Exclude infinite loops from any
23457 (pass_ch::execute): Record exits.
23459 2018-04-26 Richard Biener <rguenther@suse.de>
23461 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
23462 prologue cost vector and pass it to vect_get_load_cost.
23463 (vect_get_peeling_costs_all_drs): Likewise.
23464 (vect_peeling_hash_get_lowest_cost): Likewise.
23465 (vect_enhance_data_refs_alignment): Likewise.
23467 2018-04-26 Richard Biener <rguenther@suse.de>
23469 PR middle-end/85450
23470 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
23471 checking of integer<->pointer conversions.
23472 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
23473 sign-/zero-extending pointer types.
23474 (expand_omp_for_static_chunk): Likewise.
23476 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
23477 Jean Lee <xiaoyur347@gmail.com>
23479 * config/mips/mips.c (mips_asan_shadow_offset): New function.
23480 (TARGET_ASAN_SHADOW_OFFSET): Define.
23481 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
23482 true for -fsanitize=address.
23484 2018-04-25 Mark Wielaard <mark@klomp.org>
23486 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
23489 2018-04-25 Jakub Jelinek <jakub@redhat.com>
23491 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
23492 than "alu", remove explicit "memory" and "imm_disp" attributes.
23493 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
23495 PR middle-end/85414
23496 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
23497 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
23498 gen_lowpart_no_emit.
23500 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
23503 * config/i386/i386.c (ix86_expand_builtin): Change memory
23504 operand to XI, extend p0 to Pmode.
23505 * config/i386/i386.md: Change unspec volatile and operand
23506 1 mode to XI, change operand 0 mode to P.
23508 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
23510 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
23511 GET_MODE_MASK before any checking.
23512 (nds32_can_use_bset_p): Likewise.
23513 (nds32_can_use_btgl_p): Likewise.
23515 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
23517 * config/nds32/nds32-doubleword.md: New define_split pattern for
23518 illegal register number.
23520 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
23522 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
23524 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
23526 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
23528 2018-04-25 Richard Biener <rguenther@suse.de>
23530 * lto-streamer.h (LTO_major_version): Bump to 8.
23532 2018-04-25 Jakub Jelinek <jakub@redhat.com>
23534 * BASE-VER: Set to 9.0.0.
23536 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
23538 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
23539 in __abskf2 and __powikf2.
23541 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23544 * config/aarch64/constraints.md (Usg, Usj): New constraints.
23545 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
23546 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
23547 Use the above on operand 2. Reindent.
23548 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
23550 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
23553 * common/config/i386/i386-common.c (ix86_handle_option): Don't
23555 * config/i386/i386.opt (mcet): Removed.
23556 * doc/install.texi: Remove -mcet documentation.
23557 * doc/invoke.texi: Likewise.
23559 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
23562 * doc/install.texi: Remove -mcet from bootstrap-cet.
23564 2018-04-24 Jakub Jelinek <jakub@redhat.com>
23567 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
23568 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
23572 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
23573 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
23574 containing a CONST_VECTOR.
23576 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
23578 * doc/install.texi: Update newlib dependency for nvptx.
23580 2018-04-24 Jakub Jelinek <jakub@redhat.com>
23583 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
23584 instead of INTVAL when shifting x left.
23586 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
23588 PR tree-optimization/85478
23589 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
23590 vect_grouped_store_supported for single element vectors.
23592 2018-04-24 Richard Biener <rguenther@suse.de>
23595 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
23596 load cost increase to the case of non-constant step.
23598 2018-04-24 Jakub Jelinek <jakub@redhat.com>
23601 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
23602 destination if any_malformed_asm.
23604 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
23606 PR middle-end/85496
23607 * expr.c (store_field): In the bitfield case, if the value comes from
23608 a function call and is returned in registers by means of a PARALLEL,
23609 do not change the mode of the temporary unless BLKmode and VOIDmode.
23611 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
23613 PR rtl-optimization/85423
23614 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
23615 dependencies to debug insns when the previous insn is non-debug.
23617 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
23619 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
23620 enums into a single definition.
23621 (fls): Fix predicates and printing.
23624 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
23626 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
23627 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
23628 and short u6 immediate.
23629 (check_if_valid_sleep_operand): Remove.
23630 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
23632 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
23634 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
23635 flag_always_save_lp condition.
23636 * config/nds32/nds32.opt (malways-save-lp): New option.
23638 2018-04-22 Shiva Chen <shiva0217@gmail.com>
23640 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
23641 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
23642 * config/nds32/nds32.h
23643 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
23644 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
23646 2018-04-22 Shiva Chen <shiva0217@gmail.com>
23648 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
23649 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
23651 2018-04-22 Shiva Chen <shiva0217@gmail.com>
23652 Chung-Ju Wu <jasonwucj@gmail.com>
23654 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
23656 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
23657 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
23659 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
23661 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
23663 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
23665 * config/nds32/nds32-protos.h (nds32_data_alignment,
23666 nds32_local_alignment): Declare.
23667 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
23668 nds32_local_alignment): New functions.
23669 (TARGET_CONSTANT_ALIGNMENT): Define.
23670 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
23672 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
23674 * config/nds32/nds32.c
23675 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
23676 (TARGET_MODES_TIEABLE_P): Likewise.
23678 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
23680 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
23681 level Ofast and Og.
23683 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
23684 Chung-Ju Wu <jasonwucj@gmail.com>
23686 * config/nds32/constants.md (unspec_volatile_element): Add enum values
23687 for unaligned access.
23688 * config/nds32/nds32-intrinsic.c: Implementation of expanding
23690 * config/nds32/nds32-intrinsic.md: Likewise.
23691 * config/nds32/nds32_intrinsic.h: Likewise.
23692 * config/nds32/nds32.h (nds32_builtins): Likewise.
23693 * config/nds32/nds32.opt (munaligned-access): New option.
23694 * config/nds32/nds32.c (nds32_asm_file_start): Display
23695 flag_unaligned_access status.
23697 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
23699 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
23700 -mno-relax is present.
23701 * config/riscv/linux.h (LINK_SPEC): Ditto.
23703 2018-04-20 Martin Sebor <msebor@redhat.com>
23706 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
23708 (gimple_fold_builtin_stxcpy_chk): Same.
23709 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
23711 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
23714 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
23715 __powikf2 when long double is IEEE 128-bit.
23717 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
23719 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
23720 step to make sure stack always aligned.
23722 2018-04-20 Carl Love <cel@us.ibm.com>
23725 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
23726 size check for arg0.
23728 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
23729 Tom de Vries <tom@codesourcery.com>
23732 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
23733 Emit insns for calls too.
23734 (nvptx_find_par): Always look for worker-level predecessor insn.
23735 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
23737 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
23738 (nvptx_process_pars): Propagate frames for calls.
23740 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
23743 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
23745 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
23746 (ix86_handle_option): Don't handle OPT_mibt.
23747 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
23749 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
23751 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
23752 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
23753 (ix86_target_macros): Define __CET__ with flag_cf_protection
23754 for -fcf-protection.
23755 * config/i386/i386.c (isa2_opts): Remove -mibt.
23756 * config/i386/i386.h (TARGET_IBT): Removed.
23757 (TARGET_IBT_P): Likewise.
23758 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
23759 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
23760 * config/i386/i386.opt (mcet): Update help message.
23761 (mshstk): Likewise.
23763 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
23764 -mcet as an alias for -mshstk.
23766 2018-04-20 Richard Biener <rguenther@suse.de>
23768 PR middle-end/85475
23769 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
23770 complexity by forcing a single use of the multiply operand.
23772 2018-04-20 Martin Jambor <mjambor@suse.cz>
23775 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
23776 recursion dependency to only apply to non-clones.
23778 2018-04-20 Martin Jambor <mjambor@suse.cz>
23781 * ipa-cp.c (create_specialized_node): Check that clones of
23782 self-recursive edges exist during IPA-CP.
23784 2018-04-19 Toon Moene <toon@moene.org>
23786 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
23789 2018-04-19 Jakub Jelinek <jakub@redhat.com>
23791 PR tree-optimization/85467
23792 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
23793 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
23794 VECTOR_CST element to type.
23796 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
23799 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
23800 * config/i386/i386.md (builtin_setjmp_setup): Removed.
23801 (builtin_longjmp): Likewise.
23802 (save_stack_nonlocal): New pattern.
23803 (restore_stack_nonlocal): Likewise.
23805 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
23808 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
23809 Replace ASM_OUTPUT_LABEL with fprintf.
23811 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
23814 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
23815 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
23816 * config/i386/i386-c.c (ix86_target_macros_internal): Also
23817 define __IBT__ and __SHSTK__ for -fcf-protection.
23818 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
23820 (ix86_trampoline_init): Likewise.
23821 (x86_output_mi_thunk): Likewise.
23822 (ix86_notrack_prefixed_insn_p): Likewise.
23823 (ix86_option_override_internal): Don't disallow -fcf-protection.
23824 * config/i386/i386.md (rdssp<mode>): Also enable for
23826 (incssp<mode>): Likewise.
23827 (nop_endbr): Likewise.
23828 * config/i386/i386.opt (mcet): Change help message to built-in
23831 (mshstk): Likewise.
23832 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
23833 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
23834 enable CET built-in functions.
23836 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
23838 * common/config/i386/i386-common.c
23839 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
23840 OPTION_MASK_ISA_MOVDIRI_UNSET,
23841 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
23842 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
23843 * config.gcc (movdirintrin.h): New header.
23844 * config/i386/cpuid.h (bit_MOVDIRI,
23845 bit_MOVDIR64B): New bits.
23846 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
23848 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
23849 (VOID, PVOID, PCVOID)): New function types.
23850 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
23851 __builtin_ia32_directstoreu_u64,
23852 __builtin_ia32_movdir64b): New builtins.
23853 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
23854 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
23856 (ix86_valid_target_attribute_inner_p): Ditto.
23857 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
23858 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
23859 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
23860 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
23861 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
23862 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
23863 (movdiri<mode>, movdir64b_<mode>): New.
23864 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
23865 * config/i386/immintrin.h: Include movdirintrin.h.
23866 * config/i386/movdirintrin.h: New file.
23867 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
23869 2018-04-19 Richard Biener <rguenther@suse.de>
23871 PR middle-end/85455
23872 * cfg.c (clear_bb_flags): When loop state says we have
23873 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
23875 2018-04-19 Richard Biener <rguenther@suse.de>
23877 PR tree-optimization/84737
23878 * tree-vect-data-refs.c (vect_copy_ref_info): New function
23879 copying restrict info.
23880 (vect_setup_realignment): Use it.
23881 * tree-vectorizer.h (vect_copy_ref_info): Declare.
23882 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
23883 the first DR to all generated stores.
23884 (vectorizable_load): Likewise for loads.
23886 2018-04-19 Jakub Jelinek <jakub@redhat.com>
23888 PR tree-optimization/85446
23889 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
23890 the integral and pointer types to have the same precision.
23892 * doc/install.texi: Document --disable-cet being the default and
23895 2018-04-18 Martin Liska <mliska@suse.cz>
23897 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
23900 2018-04-18 Martin Liska <mliska@suse.cz>
23903 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
23906 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
23907 arguments if they are comparable.
23909 2018-04-18 Martin Liska <mliska@suse.cz>
23912 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
23915 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
23918 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
23921 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
23922 ENDBR after calling __morestack.
23924 2018-04-18 David Malcolm <dmalcolm@redhat.com>
23927 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
23928 by using gcc_base_ver to generate a gcc_driver_version, and use
23929 it when generating GCC_DRIVER_NAME.
23930 * configure: Regenerate.
23932 2018-04-18 Jakub Jelinek <jakub@redhat.com>
23935 * config.gcc: Obsolete powerpc*-*-*spe*.
23937 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23940 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
23941 (stabstr_D): Change type of unum from unsigned int to
23942 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
23945 2018-04-17 Jim Wilson <jimw@sifive.com>
23948 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
23949 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
23950 Set arg_pointer_offset after using pretend_args_size.
23952 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23954 PR rtl-optimization/85431
23955 * dse.c (record_store): Ignore zero width stores.
23958 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
23959 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
23960 __builtin_stack_restore rather than after it.
23961 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
23962 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
23963 argument instead of virtual_dynamic_stack_rtx.
23965 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
23967 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
23969 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
23970 Add note to error message to explain internal mapping of overloaded
23971 built-in function name to non-overloaded built-in function name.
23972 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
23975 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
23978 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
23979 where the inputs overlap with the output.
23981 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23983 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
23984 (=v, v) alternative and explicit "memory" attribute.
23985 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
23986 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
23988 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
23989 "sselog1" type instead of "sselog".
23990 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
23991 "sselog". Remove explicit "memory" attribute.
23992 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
23993 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
23995 (vec_extract_hi_v32hi): Merge all alternatives into one, use
23996 "sselog1" type instead of "sselog". Remove explicit "memory"
23998 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
23999 use "sselog1" type instead of "sselog". Remove explicit "memory"
24001 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
24002 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
24004 (vec_extract_hi_v64qi): Merge all alternatives into one, use
24005 "sselog1" type instead of "sselog". Remove explicit "memory"
24007 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
24008 use "sselog1" type instead of "sselog". Remove explicit "memory"
24012 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
24014 PR middle-end/85414
24015 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
24018 2018-04-17 Martin Jambor <mjambor@suse.cz>
24021 * ipa-cp.c (create_specialized_node): Call
24022 expand_all_artificial_thunks if necessary.
24024 2018-04-17 Martin Liska <mliska@suse.cz>
24027 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
24028 in message, remote space in between '_G' and '('.
24030 2018-04-17 Jakub Jelinek <jakub@redhat.com>
24033 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
24034 avx512f_vmcmp<mode>3<round_saeonly_name>,
24035 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
24036 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
24037 avx512f_rndscale<mode><round_saeonly_name>,
24038 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
24039 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
24040 Use %<iptr>2 instead of %2 for -masm=intel.
24041 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
24042 avx512f_vcvttss2usi<round_saeonly_name>,
24043 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
24045 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
24046 avx512f_vcvttsd2usi<round_saeonly_name>,
24047 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
24048 Use %q1 instead of %1 for -masm=intel.
24049 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
24050 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
24051 of %3 for -masm=intel.
24052 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
24054 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
24056 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
24058 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
24059 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
24061 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
24063 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
24064 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
24065 %g1 and one with %0 and %1.
24066 (avx512er_vmrcp28<mode><round_saeonly_name>,
24067 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
24068 %1 for -masm=intel.
24069 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
24070 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
24071 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
24072 of %0 and %{%4%} for -masm=intel.
24073 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
24074 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
24075 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
24076 order of %0 and %{%5%}%{z%} for -masm=intel.
24078 2018-04-17 Jan Hubicka <jh@suse.cz>
24081 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
24083 2018-04-17 Martin Liska <mliska@suse.cz>
24086 * multiple_target.c (create_dispatcher_calls): Set apostrophes
24087 for target_clone error message. Make default implementation
24088 clone to be a local declaration.
24089 (separate_attrs): Add new argument and check for an empty
24091 (expand_target_clones): Handle it.
24092 (ipa_target_clone): Make redirection just for target_clones
24095 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
24096 Tom de Vries <tom@codesourcery.com>
24098 PR middle-end/84955
24099 * omp-expand.c (expand_oacc_for): Add dummy false branch for
24100 tiled basic blocks without omp continue statements.
24102 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
24105 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
24106 vec_extract expression as having side effects to make sure it gets
24109 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
24112 * config/i386/i386.c (get_builtin_code_for_version): Check
24115 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
24118 * config.gcc: Support "skylake".
24119 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
24121 * config/i386/i386.c (m_SKYLAKE): Define.
24122 (processor_target_table): Add "skylake".
24123 (ix86_option_override_internal): Add "skylake".
24124 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
24125 PROCESSOR_CANNONLAKE.
24126 (get_builtin_code_for_version): Fix priority for
24127 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
24128 PROCESSOR_SKYLAKE-AVX512.
24129 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
24130 (processor_type): Add PROCESSOR_SKYLAKE.
24132 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
24133 Jason Merrill <jason@redhat.com>
24136 * convert.c (convert_to_integer_1): Use direct recursion for
24137 enumeral types and types with a precision less than the number
24138 of bits in their mode.
24140 2018-04-16 Julia Koval <julia.koval@intel.com>
24143 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
24144 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
24146 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
24149 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
24150 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
24151 and -mno-direct-move.
24153 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
24156 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
24157 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
24158 Ensure negative shifts result in {0}.
24160 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
24162 PR rtl-optimization/79916
24163 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
24164 regs (if any) to define how to gnerate SD moves when LRA is in
24167 2018-04-13 Jakub Jelinek <jakub@redhat.com>
24169 PR rtl-optimization/85393
24170 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
24171 * except.c (expand_dw2_landing_pad_for_region): Make static.
24172 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
24173 a label and unconditional jump to old_bb, rather than
24174 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
24177 PR rtl-optimization/85376
24178 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
24179 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
24180 instead of a specific value.
24182 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
24183 Bin Cheng <bin.cheng@arm.com>
24185 PR tree-optimization/82965
24186 PR tree-optimization/83991
24187 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
24188 by_profile_only parameter.
24189 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
24190 information if the loop was predicted to iterate too many times.
24191 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
24193 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
24196 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
24199 2018-04-13 Martin Liska <mliska@suse.cz>
24200 Jakub Jelinek <jakub@redhat.com>
24202 PR middle-end/81657
24203 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
24204 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
24205 * builtins.c (expand_builtin_memory_copy_args): Use
24206 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
24207 handle dest_addr == pc_rtx.
24209 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
24212 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
24213 asked to not generate direct moves.
24214 (fix_trunc<mode>si2_stfiwx): Similar.
24215 (fix_trunc<mode>si2_internal): Similar.
24217 2018-04-12 Jakub Jelinek <jakub@redhat.com>
24220 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
24221 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
24222 lookup if dest in some wider mode is known to be const0_rtx and
24223 if so, record permanent equivalence for it to be ZERO_EXTEND of
24224 the narrower mode destination.
24226 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
24228 * lto-streamer-out.c (output_function): Revert 259346.
24229 * omp-expand.c (expand_oacc_for): Likewise.
24231 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
24233 PR rtl-optimization/85354
24234 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
24235 * sel-sched.c (sel_global_init): ... here.
24237 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
24240 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
24241 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
24242 mode for PE-COFF targets.
24243 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
24244 (i386_pe_asm_lto_end): Likewise.
24245 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
24246 (TARGET_ASM_LTO_END): Likewise.
24247 * config/i386/winnt.c (saved_debug_info_level): New static variable.
24248 (i386_pe_asm_lto_start): New function.
24249 (i386_pe_asm_lto_end): Likewise.
24251 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
24252 Richard Biener <rguenther@suse.de>
24254 PR middle-end/84955
24255 * lto-streamer-out.c (output_function): Fix CFG loop state before
24257 * omp-expand.c (expand_oacc_for): Handle calls to internal
24258 functions like regular functions.
24260 2018-04-12 Richard Biener <rguenther@suse.de>
24263 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
24264 for the early LTO debug to properly generate references to it
24265 during DIE emission. Do not re-use that for the skeleton for
24267 (dwarf2out_early_finish): Likewise.
24269 2018-04-12 Jakub Jelinek <jakub@redhat.com>
24272 * config/i386/sse.md
24273 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
24274 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
24275 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
24276 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
24277 and output is a reg, avoid creating invalid lowpart subreg, but
24278 instead split into a 512-bit move. Don't split if not AVX512VL,
24279 input is xmm16+ reg and output is a mem.
24280 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
24281 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
24282 xmm16+ reg and output is a mem.
24284 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24286 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
24287 also for flag_dwarf2_cfi_asm.
24289 2018-04-12 Jakub Jelinek <jakub@redhat.com>
24291 PR rtl-optimization/85342
24292 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
24293 a bool scalar var inside of the loop instead. Don't try to update
24294 recog_data.operand after failed apply_change_group.
24296 2018-04-12 Tom de Vries <tom@codesourcery.com>
24299 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
24300 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
24301 array with flexible array member as array without given dimension.
24302 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
24303 argument for undefined param to true.
24305 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
24308 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
24309 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
24310 from PowerPC section.
24311 * config/rs6000/sysv4.opt (mcall-): Improve help text.
24312 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
24313 help text that is too long.
24314 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
24315 help text that is too long.
24316 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
24317 help text that is too long.
24319 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
24321 * config/alpha/alpha.md (stack_probe_internal): Rename
24322 from "probe_stack". Update all callers.
24324 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
24326 PR rtl-optimization/84566
24327 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
24328 sched_macro_fuse_insns.
24330 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
24333 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
24334 (compute_block_dependences): ... from here.
24336 2018-04-11 Jakub Jelinek <jakub@redhat.com>
24338 PR tree-optimization/85331
24339 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
24340 from int to HOST_WIDE_INT.
24342 2018-04-11 Martin Jambor <mjambor@suse.cz>
24345 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
24346 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
24347 not the same as the source val.
24348 (cgraph_edge_brings_value_p): New parameter.
24349 (gather_edges_for_value): Pass destination value to
24350 cgraph_edge_brings_value_p.
24351 (perhaps_add_new_callers): Likewise.
24352 (get_info_about_necessary_edges): Likewise and exclude values brought
24353 only by self-recursive edges.
24354 (create_specialized_node): Redirect only clones of self-calling edges.
24355 (+self_recursive_pass_through_p): New function.
24356 (find_more_scalar_values_for_callers_subset): Use it.
24357 (find_aggregate_values_for_callers_subset): Likewise.
24358 (known_aggs_to_agg_replacement_list): Removed.
24359 (decide_whether_version_node): Re-calculate known constants for all
24360 remaining context clones.
24362 2018-04-11 Richard Biener <rguenther@suse.de>
24365 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
24366 from early DWARF output.
24367 (dwarf2out_early_finish): Output line info unconditionally into
24368 early DWARF and add reference to it.
24370 2018-04-11 Jakub Jelinek <jakub@redhat.com>
24373 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
24374 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
24375 other than V2DFmode using iptr mode attribute.
24376 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
24378 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
24380 PR rtl-optimization/84659
24381 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
24383 2018-04-11 Jakub Jelinek <jakub@redhat.com>
24386 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
24388 (output_loc_list): Pass address of a dummy size variable even in the
24389 locview handling loop.
24390 (index_location_lists): Add comment on why skip_loc_list_entry can't
24393 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
24396 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
24399 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
24402 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
24403 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
24404 and -mstring-compare-inline-limit.
24406 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
24409 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
24410 for stack clash protection in a register whenever we need it to be in
24413 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
24415 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
24416 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
24418 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
24421 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
24423 (mlong-double-): Ditto.
24424 * config/rs6000/sysv4.opt (msdata=): Ditto.
24425 (mtls-size=): Ditto.
24427 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24429 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
24430 erroneous entries for
24431 "vector int vec_ldl (int, long int *)", and
24432 "vector unsigned int vec_ldl (int, unsigned long int *)".
24433 Add comments and entries for
24434 "vector bool char vec_ldl (int, bool char *)",
24435 "vector bool short vec_ldl (int, bool short *)",
24436 "vector bool int vec_ldl (int, bool int *)",
24437 "vector bool long long vec_ldl (int, bool long long *)",
24438 "vector pixel vec_ldl (int, pixel *)",
24439 "vector long long vec_ldl (int, long long *)",
24440 "vector unsigned long long vec_ldl (int, unsigned long long *)".
24441 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
24442 type tree bool_long_long_type_node and correct definition of
24443 bool_V2DI_type_node to make reference to this new type tree.
24444 (rs6000_mangle_type): Replace erroneous reference to
24445 bool_long_type_node with bool_long_long_type_node.
24446 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
24447 comments to emphasize sign distinctions for char and int types and
24448 replace RS6000_BTI_bool_long constant with
24449 RS6000_BTI_bool_long_long constant. Also add comment to restrict
24450 use of RS6000_BTI_pixel.
24451 (bool_long_type_node): Remove this macro definition.
24452 (bool_long_long_type_node): New macro definition
24454 2018-04-10 Jakub Jelinek <jakub@redhat.com>
24456 PR rtl-optimization/85300
24457 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
24458 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
24459 simplify_unary_operation fails.
24461 2018-04-10 Martin Liska <mliska@suse.cz>
24463 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
24464 cgraph_edge and ipa_ref.
24466 2018-04-10 Jakub Jelinek <jakub@redhat.com>
24470 * config/i386/sse.md
24471 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
24472 computation of the VEC_MERGE selector from mask.
24473 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
24474 Fix decoding of the VEC_MERGE selector into mask.
24476 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
24478 PR tree-optimization/85286
24479 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
24481 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
24483 * final.c (final_1): Set insn_last_address as well as
24484 insn_current_address.
24486 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24489 * explow.c (emit_stack_probe): Call validize_mem on memory location
24490 before passing it to gen_probe_stack. Create address operand and
24491 legitimize it for the probe_stack_address case.
24493 2018-04-09 Jan Hubicka <jh@suse.cz>
24496 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
24497 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
24498 * tree.c (free_lang_data_in_type): Fix handling of binfos;
24500 (free_lang_data): Rebuild type inheritance graph.
24502 2018-04-09 Martin Sebor <msebor@redhat.com>
24504 * invoke.texi (-finline-small-functions): Mention other optimization
24506 (-findirect-inlining, -fpartial-inlining): Same.
24507 (-finline-functions-called-once): Same.
24508 (-freorder-blocks-and-partition): Same.
24510 2018-04-09 Jan Hubicka <jh@suse.cz>
24513 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
24514 jumps; choose last target that matches the criteria (i.e.
24515 no partition changes for non-crossing jumps).
24516 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
24517 support for redirecting crossing jumps to non-crossing.
24519 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
24521 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
24522 also for naked functions.
24524 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
24526 * config/arc/arc.md (add_shift): New pattern.
24527 (add_shift2): Likewise.
24528 (sub_shift): Likewise.
24529 (sub_shift_cmp0_noout): Likewise.
24530 (compare_si_ashiftsi): Likewise.
24531 (xbfu_cmp0_noout): New combine pattern.
24532 (xbfu_cmp0"): Likewise.
24533 (movsi_set_cc_insn): Place the predicable variant first.
24534 (commutative_binary_cmp0_noout): Remove clobber.
24535 (commutative_binary_cmp0): New pattern.
24536 (noncommutative_binary_cmp0): Likewise.
24537 (noncommutative_binary_cmp0_noout): Likewise.
24538 (noncommutative_binary_comparison_result_used): Removed.
24539 (rsub_cmp0): New pattern.
24540 (rsub_cmp0_noout): Likewise.
24541 (extzvsi): Changed, keep only meaningful variants.
24542 (SQH, SEZ): New iterators.
24543 (SQH_postfix): New mode attribute.
24544 (SEZ_prefix): New code attribute.
24545 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
24546 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
24547 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
24548 of numerical value.
24549 (noncommutative_operator): Check the availability of barrel
24552 2018-04-09 Richard Biener <rguenther@suse.de>
24554 PR tree-optimization/85284
24555 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
24556 Only use the niter constraining form of simple_iv when the exit
24557 is always executed.
24559 2018-04-09 Tom de Vries <tom@codesourcery.com>
24562 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
24563 (define_expand "*memory_barrier"): New define_expand.
24564 (define_insn "memory_barrier"): New insn.
24566 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
24568 PR rtl-optimization/80463
24569 PR rtl-optimization/83972
24570 PR rtl-optimization/83480
24572 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
24573 correct producer for the insn.
24574 (tidy_control_flow): Fixup seqnos in case of debug insns.
24576 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
24578 PR rtl-optimization/83913
24580 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
24581 different sched-times when merging exprs.
24583 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
24585 PR rtl-optimization/83962
24587 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
24588 tidy_fallthru_edge and tidy_control_flow.
24590 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
24592 PR rtl-optimization/83530
24594 * sel-sched.c (force_next_insn): New global variable.
24595 (remove_insn_for_debug): When force_next_insn is true, also leave only
24596 next insn in the ready list.
24597 (sel_sched_region): When the region wasn't scheduled, make another pass
24598 over it with force_next_insn set to 1.
24600 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
24602 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
24604 * config/nds32/constants.md (unspec_volatile_element): Add enum values
24605 for interrupt control.
24606 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
24607 functions for interrupt control.
24608 * config/nds32/nds32-intrinsic.md: Likewise.
24609 * config/nds32/nds32_intrinsic.h: Likewise.
24610 * config/nds32/nds32.h (nds32_builtins): Likewise.
24612 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
24614 * config/nds32/nds32.c (nds32_init_machine_status,
24615 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
24616 strict_aligned_p field.
24617 (nds32_expand_to_rtl_hook): New function.
24618 (TARGET_EXPAND_TO_RTL_HOOK): Define.
24619 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
24621 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
24622 Chung-Ju Wu <jasonwucj@gmail.com>
24624 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
24625 * config/nds32/nds32-n7.md: New file.
24626 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
24627 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
24629 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
24630 * config/nds32/nds32.md (pipeline_model): Add n7.
24631 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
24632 * config/nds32/pipelines.md: Include n7 settings.
24634 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
24635 Chung-Ju Wu <jasonwucj@gmail.com>
24637 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
24638 * config/nds32/nds32-e8.md: New file.
24639 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
24640 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
24642 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
24643 * config/nds32/nds32.md (pipeline_model): Add e8.
24644 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
24645 * config/nds32/pipelines.md: Include e8 settings.
24647 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
24648 Chung-Ju Wu <jasonwucj@gmail.com>
24650 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
24651 * config/nds32/nds32-n8.md: New file.
24652 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
24653 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
24655 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
24656 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
24657 * config/nds32/nds32.md (pipeline_model): Add n8.
24658 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
24659 * config/nds32/pipelines.md: Include n8 settings.
24661 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
24662 Chung-Ju Wu <jasonwucj@gmail.com>
24664 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
24665 * config/nds32/nds32-n9-2r1w.md: New file.
24666 * config/nds32/nds32-n9-3r2w.md: New file.
24667 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
24668 nds32_register_ports): New or modify for cpu n9.
24669 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
24671 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
24672 * config/nds32/nds32-utils.c: New file.
24673 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
24674 TARGET_MUL_SLOW): Define.
24675 * config/nds32/nds32.md (pipeline_model): New attribute.
24676 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
24677 New options that support cpu n9.
24678 * config/nds32/pipelines.md: Include n9 settings.
24679 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
24681 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
24683 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
24684 information if necessary.
24685 (output_cond_branch_compare_zero): Likewise.
24686 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
24687 (nds32_target_alignment): Refine for alignment.
24688 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
24689 (FUNCTION_BOUNDARY): Modify.
24690 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
24692 * config/nds32/nds32.opt (malways-align, malign-functions): New.
24694 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
24696 * config/nds32/constants.md (unspec_volatile_element): Add values for
24697 TLB operation and data prefetch.
24698 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
24699 functions for TLB operation and data prefetch.
24700 * config/nds32/nds32-intrinsic.md: Likewise.
24701 * config/nds32/nds32_intrinsic.h: Likewise.
24702 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
24703 (nds32_print_operand): Likewise.
24704 * config/nds32/nds32.h (nds32_builtins): Likewise.
24706 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
24707 Andrew Pinski <pinsika@gcc.gnu.org>
24709 PR middle-end/82976
24710 * match.pd: Use constant_boolean_node of correct type instead of
24711 boolean_true_node or boolean_false_node for simplifying
24712 pointer comparisons to zero.
24714 2018-04-07 Jakub Jelinek <jakub@redhat.com>
24716 PR tree-optimization/80021
24717 * tree.c (verify_type_variant): Make error call in verify_variant_match
24718 translatable and remove final full stop.
24720 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
24722 * config/nds32/constants.md (unspec_volatile_element): Add
24723 UNSPEC_VOLATILE_EH_RETURN.
24724 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
24725 nds32_output_stack_pop): Support dwarf exception handling process.
24726 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
24727 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
24728 exception handling process.
24729 (nds32_compute_stack_frame): Likewise.
24730 (nds32_return_addr_rtx): Likewise.
24731 (nds32_initial_elimination_offset): Likewise.
24732 (nds32_expand_prologue): Likewise.
24733 (nds32_expand_epilogue): Likewise.
24734 (nds32_dynamic_chain_address): New function.
24735 * config/nds32/nds32.h (machine_function): Add fields for dwarf
24736 exception handling.
24737 (DYNAMIC_CHAIN_ADDRESS): Define.
24738 (EH_RETURN_DATA_REGNO): Define.
24739 (EH_RETURN_STACKADJ_RTX): Define.
24740 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
24741 patterns for dwarf exception handling.
24743 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
24745 * config/nds32/nds32.h: Clean up obsolete macros.
24747 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24749 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
24750 Add enum values for particular instructions.
24751 * config/nds32/nds32-intrinsic.c: Implementation of expanding
24752 particular intrinsic functions.
24753 * config/nds32/nds32-intrinsic.md: Likewise.
24754 * config/nds32/nds32_intrinsic.h: Likewise.
24755 * config/nds32/nds32.h (nds32_builtins): Likewise.
24756 * config/nds32/nds32.md (type): Add pbsad and pbsada.
24757 (btst, ave): New patterns for particular instructions.
24759 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24761 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
24762 Add enum values for atomic load/store and memory sync.
24763 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
24765 * config/nds32/nds32-intrinsic.md: Likewise.
24766 * config/nds32/nds32_intrinsic.h: Likewise.
24767 * config/nds32/nds32.h (nds32_builtins): Likewise.
24769 2018-04-07 Jakub Jelinek <jakub@redhat.com>
24771 PR tree-optimization/85257
24772 * fold-const.c (native_encode_vector): If not all elts could fit
24773 and off is -1, return 0 rather than offset.
24774 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
24775 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
24776 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
24777 adjust buffer in native_interpret_expr call.
24779 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24781 * config/nds32/constants.md (unspec_volatile_element): Add cache
24782 control enum values.
24783 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
24784 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
24785 * config/nds32/nds32.c (nds32_cctl_names): New.
24786 (nds32_print_operand): Handle cache control register names.
24787 * config/nds32/nds32.h (nds32_builtins): New enum values.
24788 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
24790 * config/nds32/nds32.md (type): Add mmu.
24791 * config/nds32/pipelines.md (simple_insn): Add mmu.
24793 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
24795 * config/nds32/nds32.md (type): Remove call.
24796 * config/nds32/pipelines.md (simple_insn): Likewise.
24798 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24800 * config/nds32/constants.md (unspec_volatile_element): Add
24801 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
24802 UNSPEC_VOLATILE_FMFCFG.
24803 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
24804 description for fmfcfg and fmfcsr.
24805 (bdesc_1arg): Add fmtcsr.
24806 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
24807 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
24808 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
24809 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
24810 unspec_fmfcfg): New patterns.
24811 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
24812 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
24813 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
24814 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
24815 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
24816 __nds32__fmfcfg): Define.
24818 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24820 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
24821 intrinsic register names.
24822 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
24823 intrinsic register enum values and macros.
24825 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
24827 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
24828 for load/store addressing form.
24829 (nds32_print_operand_address): Likewise.
24831 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
24834 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
24835 based on LABEL_REF. Remove useless assertion.
24836 (pic_address_needs_scratch): Fix formatting.
24837 (sparc_legitimize_pic_address): Minor tweaks.
24838 (sparc_delegitimize_address): Adjust assertion accordingly.
24839 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
24840 into symbolic_operand.
24841 (movsi_high_pic_label_ref): Likewise.
24842 (movsi_lo_sum_pic_label_ref): Likewise.
24843 (movdi_pic_label_ref): Likewise.
24844 (movdi_high_pic_label_ref): Likewise.
24845 (movdi_lo_sum_pic_label_ref): Likewise.
24847 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
24849 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
24850 custom LIB_SPEC setup.
24852 2018-04-06 Ruslan Bukin <br@bsdpad.com>
24853 Kito Cheng <kito.cheng@gmail.com>
24855 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
24856 * config/riscv/freebsd.h: New.
24858 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
24860 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
24861 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
24864 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
24865 Kito Cheng <kito.cheng@gmail.com>
24867 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
24868 nds32_output_call, nds32_symbol_binds_local_p): New functions.
24869 * config/nds32/nds32-protos.h (nds32_output_call,
24870 nds32_output_return): Declare.
24871 * config/nds32/nds32.md: Refine all the call and return patterns.
24873 2018-04-06 Jakub Jelinek <jakub@redhat.com>
24876 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
24877 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
24879 PR rtl-optimization/84872
24880 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
24881 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
24882 EDGE_CROSSING edge.
24884 2018-04-06 Tamar Christina <tamar.christina@arm.com>
24886 * expr.c (copy_blkmode_to_reg): Revert 254862.
24887 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
24889 2018-04-06 Richard Biener <rguenther@suse.de>
24891 PR middle-end/85244
24892 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
24893 after seeing a component reference with an adjacent field. Treat
24894 refs to arrays at struct end of external decls similar to
24895 refs to unconstrained commons.
24897 2018-04-06 Jakub Jelinek <jakub@redhat.com>
24900 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
24901 look through SAVE_EXPRs with non-side-effects argument. Adjust
24903 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
24906 2018-04-06 Richard Biener <rguenther@suse.de>
24908 PR middle-end/85180
24909 * alias.c (find_base_term): New wrapper around find_base_term
24910 unwinding CSELIB_VAL_PTR changes.
24911 (find_base_term): Do not restore CSELIB_VAL_PTR during the
24914 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24916 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
24918 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
24919 constant definitions.
24920 ("nop"): lr 0,0 -> nopr r0
24921 ("nop_lr0", "nop_lr1"): New insn definitions.
24923 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
24925 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
24926 NDS32_V3PUSH_AVAILABLE_P macro.
24928 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
24929 Chung-Ju Wu <jasonwucj@gmail.com>
24931 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
24932 (nds32*-*-*): Add float and fpu_config into supported_defaults.
24933 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
24934 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
24935 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
24936 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
24937 * config/nds32/constraints.md: New constraints and checking for hard
24938 float configuration.
24939 * config/nds32/iterators.md: New mode iterator and attribute for hard
24940 float configuration.
24941 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
24943 * config/nds32/nds32-fpu.md: New file.
24944 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
24945 deal with hard float code generation.
24946 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
24948 (abi_type, float_reg_number): New enum type.
24949 * config/nds32/nds32-predicates.c: New predicates for hard float.
24950 * config/nds32/nds32-protos.h: Declare functions for hard float.
24951 * config/nds32/nds32.c: Implementation for hard float configuration.
24952 * config/nds32/nds32.h: Definitions for hard float configuration.
24953 * config/nds32/nds32.md: Include hard float machine description and
24954 modify patterns for hard float configuration.
24955 * config/nds32/nds32.opt: New options for hard float configuration.
24956 * config/nds32/predicates.md: New predicates for hard float
24959 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24961 * common/config/nds32/nds32-common.c
24962 (nds32_option_optimization_table): Enable -mreleax-hint by default.
24964 2018-04-05 Jakub Jelinek <jakub@redhat.com>
24966 PR middle-end/85195
24967 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
24968 CONSTRUCTOR_ELT (ctor, ...)->value.
24970 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
24973 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
24975 2018-04-05 Tom de Vries <tom@codesourcery.com>
24978 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
24981 2018-04-05 Shiva Chen <shiva0217@gmail.com>
24982 Kito Cheng <kito.cheng@gmail.com>
24984 * config/nds32/constraints.md (U33): Fine-tune checking condition.
24985 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
24986 * config/nds32/nds32.h (nds32_16bit_address_type): Add
24987 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
24989 2018-04-05 Shiva Chen <shiva0217@gmail.com>
24990 Kito Cheng <kito.cheng@gmail.com>
24992 * config/nds32/constraints.md (Ufe): New memory constraint.
24993 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
24994 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
24995 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
24997 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
24998 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
25000 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
25002 * config/nds32/nds32.md: Use optimize_size in the condition for
25003 alu-shift instructions.
25005 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
25007 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
25009 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
25011 * config/nds32/nds32.md (negsi2): Refine pattern.
25013 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
25014 Chung-Ju Wu <jasonwucj@gmail.com>
25016 * config/nds32/iterators.md (shift_rotate): New code iterator.
25017 (shift): New code attribute.
25018 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
25019 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
25020 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
25021 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
25022 bit-wise operations.
25023 (andsi3, *andsi3): Ditto.
25024 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
25025 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
25026 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
25027 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
25028 nds32_ior_operand, nds32_xor_operand): New predicates.
25030 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
25032 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
25033 (addsi3, subsi3): ... this.
25035 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
25037 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
25039 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
25041 * config/nds32/nds32.md: Adjust indention.
25043 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
25045 * config/nds32/nds32.md (feature): New attribute.
25047 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
25049 * config/nds32/nds32.md (subtype): New attribute.
25051 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25054 * config/arm/arm-builtins.c (arm_expand_builtin): Change
25055 expansion to perform a bitwise AND of the argument followed by a
25056 boolean negation of the result.
25058 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
25060 PR rtl-optimization/84878
25061 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
25062 the basic block. Assert the use reference is not artificial and that
25063 it has an associated insn.
25065 2018-04-04 Michael Matz <matz@suse.de>
25067 * builtins.c (compute_objsize): Pass correct operand
25068 to array_at_struct_end_p.
25070 2018-04-04 Richard Biener <rguenther@suse.de>
25073 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
25074 from contexts for DINFO_LEVEL_TERSE and below.
25076 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
25078 * config/nds32/nds32-doubleword.md (move_<mode>): Require
25079 resiter_operand condition.
25080 * config/nds32/nds32.md (*move<mode>): Ditto.
25082 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
25083 Monk Chiang <sh.chiang04@gmail.com>
25085 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
25087 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
25089 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
25091 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
25092 Kito Cheng <kito.cheng@gmail.com>
25094 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
25095 nds32_cond_code_str, output_cond_branch,
25096 output_cond_branch_compare_zero, nds32_expand_cbranch,
25097 nds32_expand_cstore, nds32_expand_movcc,
25098 nds32_output_cbranchsi4_equality_zero,
25099 nds32_output_cbranchsi4_equality_reg,
25100 nds32_output_cbranchsi4_equality_reg_or_const_int,
25101 nds32_output_cbranchsi4_greater_less_zero: New functions.
25102 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
25103 nds32_expand_cstore, nds32_expand_movcc,
25104 nds32_output_cbranchsi4_equality_zero,
25105 nds32_output_cbranchsi4_equality_reg,
25106 nds32_output_cbranchsi4_equality_reg_or_const_int,
25107 nds32_output_cbranchsi4_greater_less_zero): Declare.
25108 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
25109 nds32_rimm11s_operand): New predicates.
25110 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
25111 * config/nds32/nds32.md: Rewrite all the branch and conditional move
25114 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
25116 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
25117 * config/nds32/nds32.md: Ditto.
25118 * config/nds32/pipelines.md: Ditto.
25120 2018-04-04 Richard Biener <rguenther@suse.de>
25122 PR tree-optimization/85168
25123 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
25124 propagating abnormals.
25126 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
25128 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
25130 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
25131 Kito Cheng <kito.cheng@gmail.com>
25133 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
25134 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
25135 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
25136 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
25137 * config/nds32/nds32.md (sibcall_internal): New.
25138 (sibcall_register): Remove.
25139 (sibcall_immediate): Remove.
25140 (sibcall_value_internal): New.
25141 (sibcall_value_register): Remove.
25142 (sibcall_value_immediate): Remove.
25143 * config/nds32/predicates.md (nds32_general_register_operand): New.
25144 (nds32_call_address_operand): New.
25146 2018-04-03 Jakub Jelinek <jakub@redhat.com>
25148 PR rtl-optimization/85167
25149 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
25150 bb_defs if *split_p, instead preinitialize it to NULL.
25152 PR tree-optimization/85156
25153 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
25154 evaluating the argument multiple times.
25156 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
25158 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
25160 (_mm_cvtpd_ps): Likewise.
25161 (_mm_cvttpd_epi32): Likewise.
25162 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
25163 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
25164 vector, pixel, and bool following altivec.h include.
25166 2018-04-03 Martin Sebor <msebor@redhat.com>
25168 * doc/extend.texi (Common Function Attributes): Clarify.
25169 (const attribute): Likewise.
25170 (pure attribute): Likewise.
25172 2018-04-03 Jakub Jelinek <jakub@redhat.com>
25175 * config/i386/i386.c (ix86_expand_vector_set): Use
25176 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
25178 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
25180 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
25181 instructions when changing rounding bits to preserve precision bits
25182 in the x87 control word.
25184 2018-04-03 Martin Liska <mliska@suse.cz>
25186 PR tree-optimization/82491
25187 * rtl.h (strip_offset_and_add): Replace += suboffset with
25188 poly_uint64 () + suboffset.
25190 2018-03-29 Martin Liska <mliska@suse.cz>
25191 Martin Jambor <mjambor@suse.cz>
25194 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
25195 param_type is not an integral or pointer type.
25197 2018-04-03 Richard Biener <rguenther@suse.de>
25199 * sese.h (recompute_all_dominators): Remove.
25201 2018-04-02 Martin Sebor <msebor@redhat.com>
25203 * doc/invoke.texi (-Wrestrict): Fix typos.
25205 2018-04-02 Jim Wilson <jimw@sifive.com>
25207 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
25208 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
25209 (<optab>di3, <optab>si3_extend): Likewise.
25210 (<optab>si3_mask, <optab>si3_mask_1): New.
25211 (<optab>di3_mask, <optab>di3_mask_1): New.
25212 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
25213 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
25214 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
25216 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
25218 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
25221 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
25223 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
25224 (nds32_canonicalize_comparison): New function.
25226 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
25227 Kito Cheng <kito.cheng@gmail.com>
25228 Kuan-Lin Chen <kuanlinchentw@gmail.com>
25230 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
25231 * config/nds32/constants.md (unspec_volatile_element): Add
25232 UNSPEC_VOLATILE_RELAX_GROUP.
25233 * config/nds32/nds32-relax-opt.c: New file.
25234 * config/nds32/nds32-predicates.c
25235 (nds32_symbol_load_store_p): New function.
25236 * config/nds32/nds32-protos.h
25237 (nds32_symbol_load_store_p): Declare function.
25238 (make_pass_nds32_relax_opt): Declare new rtl pass function.
25239 * config/nds32/nds32.c
25240 (nds32_register_pass): New function to register pass.
25241 (nds32_register_passes): New function to register passes.
25242 * config/nds32/nds32.md (relax_group): New pattern.
25243 * config/nds32/nds32.opt (mrelax-hint): New option.
25244 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
25246 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
25248 * config/nds32/t-nds32: Modify files dependency.
25250 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
25252 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
25253 (PROFILE_HOOK): Define its implementation.
25255 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
25257 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
25258 type and 32-bit size.
25260 2018-04-01 Jakub Jelinek <jakub@redhat.com>
25262 PR middle-end/85090
25263 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
25264 (V_128_256): New mode iterator.
25265 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
25266 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
25267 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
25269 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
25270 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
25272 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
25275 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
25276 NaN inputs correctly.
25278 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
25281 * config/rs6000/vsx.md (??r): New mode attribute.
25282 (*vsx_mov<mode>_64bit): Use it.
25283 (*vsx_mov<mode>_32bit): Likewise.
25285 2018-03-30 Martin Sebor <msebor@redhat.com>
25287 PR tree-optimization/84818
25288 * builtins.c (check_access): Use warning_n.
25290 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25293 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
25295 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
25298 2018-03-30 Julia Koval <julia.koval@intel.com>
25301 * x86-tune.def (movx, partial_reg_dependency): Enable for
25304 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
25306 PR inline-asm/84985
25307 * lra-constraints.c (process_alt_operands): Move setting
25308 this_alternative_matches below.
25310 2018-03-29 Martin Liska <mliska@suse.cz>
25313 * doc/invoke.texi: Document how LTO works with debug info.
25314 Describe auto-load support of binutils. Mention 'x86-64'
25315 as valid option value of -march option.
25317 2018-03-29 Jakub Jelinek <jakub@redhat.com>
25319 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
25322 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
25323 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
25324 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
25327 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
25330 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
25331 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
25332 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
25333 for RS6000_BTM_POWERPC64.
25334 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
25335 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
25336 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
25339 (DIVDEU): Likewise.
25341 2018-03-28 Carl Love <cel@us.ibm.com>
25344 2017-09-27 Carl Love <cel@us.ibm.com>
25346 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
25347 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
25348 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
25351 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25353 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
25354 instead of __vector bool.
25355 (_mm_max_pu8): Likewise.
25356 (_mm_min_pi16): Likewise.
25358 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
25361 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
25362 (DIVWEUO): Likewise.
25363 (DIVDEO): Likewise.
25364 (DIVDEUO): Likewise.
25365 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
25366 DIVWEUO and DIVDEUO.
25367 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
25368 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
25369 (div_extend): Likewise.
25370 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
25372 (__builtin_divweuo): Likewise.
25373 (__builtin_divdeo): Likewise.
25374 (__builtin_divdeuo): Likewise.
25376 2018-03-28 Jakub Jelinek <jakub@redhat.com>
25379 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
25380 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
25382 PR tree-optimization/82004
25383 * gimple-match-head.c (optimize_pow_to_exp): New function.
25384 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
25385 Don't fold to exp if optimize_pow_to_exp is false.
25387 2018-03-28 Martin Liska <mliska@suse.cz>
25390 * calls.c (initialize_argument_information): Fix trailing space.
25391 * common.opt: Fix typo and provide better explanation for
25392 -fsanitize-coverage option.
25393 * config/i386/i386.opt: Fix typo.
25395 2018-03-28 Jakub Jelinek <jakub@redhat.com>
25396 Martin Liska <mliska@suse.cz>
25399 * gimplify.c (asan_poison_variable): Don't do the check for
25400 gimplify_omp_ctxp here.
25401 (gimplify_decl_expr): Do it here.
25402 (gimplify_target_expr): Likewise.
25404 2018-03-28 Martin Liska <mliska@suse.cz>
25407 * config/i386/i386.c (ix86_function_arg_advance): Do not call
25408 chkp_type_bounds_count if MPX is not enabled.
25410 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
25412 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
25414 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
25417 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
25418 function to create the function decl for complex long double
25419 multiply and divide for -mabi=ieeelongdouble.
25420 (init_float128_ieee): Call it.
25422 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
25425 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
25426 -fcf-protection=branch -mibt.
25427 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
25429 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25432 * config/arm/arm.c (arm_valid_symbolic_address): Handle
25433 arm_word_relocations.
25435 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
25438 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
25439 extern array declarations.
25441 2018-03-27 Richard Biener <rguenther@suse.de>
25443 PR middle-end/84067
25444 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
25445 explicit single_use checks.
25447 2018-03-27 Richard Biener <rguenther@suse.de>
25449 PR tree-optimization/85082
25450 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
25453 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25455 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
25456 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
25457 Turn on fasynchronous-unwind-tables and funwind-tables.
25459 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
25462 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
25463 (*bmi_blsr_<mode>_ccz): Ditto.
25465 2018-03-26 Tom de Vries <tom@codesourcery.com>
25467 PR tree-optimization/85063
25468 * omp-general.c (offloading_function_p): New function. Factor out
25470 * omp-offload.c (pass_omp_target_link::gate): ... here.
25471 * omp-general.h (offloading_function_p): Declare.
25472 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
25473 with attribute omp declare target for offloading functions.
25475 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
25477 PR tree-optimization/84005
25478 * tree-data-ref.h (get_base_for_alignment): Declare.
25479 * tree-data-ref.c (get_base_for_alignment_1): New function.
25480 (get_base_for_alignment): Likewise.
25481 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25482 get_base_for_alignment to find a suitable base object, instead
25483 of always using drb->base_address.
25485 2018-03-23 Jakub Jelinek <jakub@redhat.com>
25487 PR inline-asm/85022
25488 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
25489 known size by default.
25491 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
25493 PR inline-asm/85030
25494 * lra-constraints.c (process_alt_operands): Don't match BLKmode
25495 and non BLKmode operands.
25497 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25500 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
25501 Clean up attributes.
25503 2018-03-23 Richard Biener <rguenther@suse.de>
25506 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
25507 we are going to emit early debug for LTO.
25509 2018-03-23 Jakub Jelinek <jakub@redhat.com>
25511 PR inline-asm/85034
25512 * function.c (match_asm_constraints_1): Don't optimize if input
25513 doesn't satisfy general_operand predicate for output's mode.
25515 PR inline-asm/85022
25516 * alias.c (write_dependence_p): Don't require for x_canonicalized
25517 non-VOIDmode if x has VOIDmode.
25520 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
25521 just don't try to optimize it rather than assert it never happens.
25523 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
25525 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
25526 macro expansions for definition of ST_INTERNAL_<mode> and
25527 LD_INTERNAL_<mode> builtins.
25528 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
25530 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
25532 (altivec_expand_st_builtin): Likewise.
25533 (altivec_expand_builtin): Remove calls to deleted functions.
25534 (rs6000_address_for_altivec): Delete this function.
25535 * config/rs6000/vector.md: Remove expands for
25536 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
25538 2018-03-22 Sudakshina Das <sudi.das@arm.com>
25541 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
25542 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
25543 re-computing once computed.
25544 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
25545 (arm_init_machine_status): Initialize
25546 machine->static_chain_stack_bytes.
25548 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
25551 * doc/extend.texi: Add four new prototypes for vec_ld.
25552 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
25553 definitions for more logical presentation.
25554 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
25555 entries for V1TI variants of __builtin_altivec_ld builtin.
25556 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
25557 handling of V1TI variant of LVX icode pattern.
25558 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
25559 (rs6000_gimple_fold_builtin): Likewise.
25560 (altivec_init_builtins): Add code to define
25561 __builtin_altivec_lvx_v1ti function.
25563 2018-03-22 Jakub Jelinek <jakub@redhat.com>
25565 PR inline-asm/84941
25566 * function.c (match_asm_constraints_1): Don't do the optimization
25567 if input isn't a REG, SUBREG, MEM or constant.
25569 2018-03-22 Tom de Vries <tom@codesourcery.com>
25571 PR tree-optimization/84956
25572 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
25573 bb_has_abnormal_pred.
25575 2018-03-22 Jakub Jelinek <jakub@redhat.com>
25578 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
25579 DECL_INITIAL (decl) to decl at the end.
25580 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
25581 adjust the comment.
25583 2018-03-21 Joseph Myers <joseph@codesourcery.com>
25585 * doc/extend.texi (__builtin_tgmath): Document when complex
25586 integer types are treated as _Complex _Float64.
25588 2018-03-21 Tom de Vries <tom@codesourcery.com>
25590 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
25592 2018-03-21 Jakub Jelinek <jakub@redhat.com>
25594 PR tree-optimization/84960
25595 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
25596 if it is ENTRY block, move them into single succ of ENTRY in that case.
25598 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
25600 PR tree-optimization/84811
25601 * poly-int.h (poly_span_traits): Remove the T3 parameter and
25602 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
25603 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
25604 (known_subrange_p): Update accordingly. Cast each value involved
25605 in the size comparison, rather than casting the result of the
25608 2018-03-21 Jakub Jelinek <jakub@redhat.com>
25610 PR tree-optimization/84982
25611 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
25612 by flipping the least significant bit rather than all bits from
25613 bitpos to bitpos + bitsize - 1.
25615 2018-03-21 Nathan Sidwell <nathan@acm.org>
25617 * doc/extend.texi (Deprecated Features): Remove mention of
25618 long-deleted deprecations.
25620 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25623 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
25624 * configure: Regenerate.
25626 2018-03-21 Tom de Vries <tom@codesourcery.com>
25628 PR tree-optimization/83126
25629 * tree-parloops.c (num_phis): New function.
25630 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
25632 2018-03-21 Nathan Sidwell <nathan@acm.org>
25634 * doc/extend.texi (Deprecated Features): Update deprecated flags,
25635 mention anon-struct/union members and trailing attributes.
25637 2018-03-21 Bin Cheng <bin.cheng@arm.com>
25639 PR tree-optimization/84969
25640 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
25641 builtin memset partitions if they set different rhs values.
25643 2018-03-21 Jakub Jelinek <jakub@redhat.com>
25645 PR rtl-optimization/84989
25646 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
25647 VEC_DUPLICATE with scalar result mode.
25649 2018-03-21 Martin Liska <mliska@suse.cz>
25652 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
25653 not intended return statement.
25655 2018-03-21 Martin Liska <mliska@suse.cz>
25658 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
25659 (chkp_find_bound_slots_1): Limit number of iterations.
25661 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
25664 * Minor grammar fixes for x86 options.
25666 2018-03-20 Jakub Jelinek <jakub@redhat.com>
25669 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
25670 holding REG_CFA_RESTORE notes, instead turn them into a USE.
25672 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
25675 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
25676 (altivec_lvx_<mode>_1op): Likewise.
25677 (altivec_stvx_<mode>_2op): Likewise.
25678 (altivec_stvx_<mode>_1op): Likewise.
25679 (altivec_lvx_<VM2:mode>): New define_expand.
25680 (altivec_stvx_<VM2:mode>): Likewise.
25681 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
25682 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
25683 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
25684 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
25685 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
25686 (rs6000_gen_lvx): Likewise.
25687 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
25688 (altivec_expand_stv_builtin): Likewise.
25689 (altivec_expand_builtin): Likewise.
25690 * config/rs6000/vector.md: Likewise.
25692 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25695 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
25698 2018-03-20 Richard Biener <rguenther@suse.de>
25701 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
25702 sign-conversions as zero, fall back to standard scalar_stmt
25705 2018-03-20 Martin Liska <mliska@suse.cz>
25708 * predict.c (rebuild_frequencies): Handle case when we have
25709 PROFILE_ABSENT, but flag_guess_branch_prob is false.
25711 2018-03-20 Jakub Jelinek <jakub@redhat.com>
25714 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
25715 flag_section_anchors.
25716 * varasm.c (use_blocks_for_decl_p): Remove hack for
25717 dw2_force_const_mem.
25720 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
25722 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
25723 be created, use lowpart_subreg of operands[0] rather than operands[0]
25725 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
25726 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
25727 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
25728 and n constraint instead of aarch64_shift_imm_di and Usd.
25729 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
25730 (*aarch64_<optab>_reg_minus<mode>3): ... this.
25732 2018-03-20 Sudakshina Das <sudi.das@arm.com>
25735 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
25736 to favor GPR over NEON registers.
25737 (<shift>di3_neon): Likewise.
25739 2018-03-20 Tom de Vries <tom@codesourcery.com>
25742 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
25743 (nvptx_process_pars): Emit bar.sync asap and alap.
25745 2018-03-20 Tom de Vries <tom@codesourcery.com>
25748 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
25749 seen_label if seen_label is already set.
25751 2018-03-20 Jakub Jelinek <jakub@redhat.com>
25754 * config/i386/i386.c (fold_builtin_cpu): For features above 31
25755 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
25756 Use 1U instead of 1. Formatting fixes.
25759 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
25760 instead of TREE_TYPE (s1) for the return value.
25762 2018-03-19 Jakub Jelinek <jakub@redhat.com>
25764 PR tree-optimization/84946
25765 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
25766 bitsize + bitsize in poly_uint64 rather than poly_int64.
25769 * dwarf2asm.c: Include fold-const.c.
25770 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
25771 of decl rather than decl itself.
25773 PR rtl-optimization/84643
25774 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
25776 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
25779 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
25780 calling assemble_variable.
25782 2018-03-19 Sudakshina Das <sudi.das@arm.com>
25785 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
25786 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
25788 2018-03-19 Jim Wilson <jimw@sifive.com>
25791 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
25792 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
25793 (riscv_first_stack_step): Likewise.
25794 (riscv_option_override): Use STACK_BOUNDARY instead of
25795 MIN_STACK_BOUNDARY.
25796 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
25797 MIN_STACK_BOUNDARY.
25798 (BIGGEST_ALIGNMENT): Set to 128.
25799 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
25800 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
25803 2018-03-19 Richard Biener <rguenther@suse.de>
25805 PR tree-optimization/84933
25806 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
25807 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
25809 2018-03-19 Richard Biener <rguenther@suse.de>
25811 PR tree-optimization/84859
25812 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
25813 (cond_if_else_store_replacement): Perform sinking operation on
25814 single-store BBs regardless of MAX_STORES_TO_SINK setting.
25815 Generalize what a BB with a single eligible store is.
25817 2018-03-19 Richard Biener <rguenther@suse.de>
25819 PR tree-optimization/84929
25820 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
25821 chrec_is_positive against non-chrec arg.
25823 2018-03-19 Tamar Christina <tamar.christina@arm.com>
25826 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
25828 2018-03-18 Martin Liska <mliska@suse.cz>
25830 PR rtl-optimization/84635
25831 * regrename.c (build_def_use): Use matches_mode only when
25834 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
25836 PR tree-optimization/84913
25837 * tree-vect-loop.c (vectorizable_reduction): Don't try to
25838 vectorize chains of COND_EXPRs.
25840 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
25842 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
25844 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
25846 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
25848 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
25850 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
25852 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
25853 Kito Cheng <kito.cheng@gmail.com>
25855 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
25856 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
25857 (nds32_adjust_reg_alloc_order): New function.
25858 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
25860 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
25862 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
25863 nds32_print_operand, nds32_print_operand_address): Use
25864 HOST_WIDE_INT_PRINT_DEC instead.
25866 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
25868 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
25870 2018-03-17 Jakub Jelinek <jakub@redhat.com>
25873 * config/i386/i386.c (initial_ix86_tune_features,
25874 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
25875 unsigned long long.
25876 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
25877 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
25878 rather than 1u << ix86_tune. Formatting fix.
25879 (ix86_option_override_internal): Change ix86_arch_mask from
25880 unsigned int to unsigned HOST_WIDE_INT, initialize to
25881 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
25882 (ix86_function_specific_restore): Likewise.
25884 2018-03-16 Jakub Jelinek <jakub@redhat.com>
25887 * postreload.c (reload_combine_recognize_pattern): Perform
25888 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
25889 truncate_int_for_mode the result for the destination's mode.
25892 * hsa-gen.c (mem_type_for_type): Fix comment typo.
25893 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
25895 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
25898 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
25901 * lra-assigns.c (lra_split_hard_reg_for): Don't use
25902 regno_allocno_class_array and sorted_pseudos.
25903 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
25904 insns where regno is used.
25906 2018-03-16 Martin Liska <mliska@suse.cz>
25909 * multiple_target.c (create_dispatcher_calls): Redirect
25910 reference in the symbol table.
25912 2018-03-16 Martin Liska <mliska@suse.cz>
25915 * multiple_target.c (create_dispatcher_calls): Redirect also
25918 2018-03-16 Jakub Jelinek <jakub@redhat.com>
25922 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
25923 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
25924 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
25926 2018-03-16 Julia Koval <julia.koval@intel.com>
25928 * doc/invoke.texi (Skylake Server): Add CLWB.
25929 Cannonlake): Remove CLWB.
25931 2018-03-16 Jakub Jelinek <jakub@redhat.com>
25933 PR tree-optimization/84841
25934 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
25936 (FLOAT_ONE_CONST_TYPE): Define.
25937 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
25938 (sort_by_operand_rank): Put entries with higher constant_type last
25939 rather than first to match comments.
25941 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
25943 * config/nios2/nios2.md (movsi_internal): Fix thinko in
25946 2018-03-15 Jakub Jelinek <jakub@redhat.com>
25949 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
25950 check and use address of target always.
25952 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
25955 * config/i386/i386.c (indirect_thunk_needed): Update comments.
25956 (indirect_thunk_bnd_needed): Likewise.
25957 (indirect_thunks_used): Likewise.
25958 (indirect_thunks_bnd_used): Likewise.
25959 (indirect_return_needed): New.
25960 (indirect_return_bnd_needed): Likewise.
25961 (output_indirect_thunk_function): Add a bool argument for
25963 (output_indirect_thunk_function): Don't generate alias for
25964 function return thunk.
25965 (ix86_code_end): Call output_indirect_thunk_function to generate
25966 function return thunks.
25967 (ix86_output_function_return): Set indirect_return_bnd_needed
25968 and indirect_return_needed instead of indirect_thunk_bnd_needed
25969 and indirect_thunk_needed.
25971 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
25973 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
25974 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
25975 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
25977 2018-03-15 David Malcolm <dmalcolm@redhat.com>
25978 Paul Hua <paul.hua.gm@gmail.com>
25981 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
25983 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
25985 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
25986 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
25987 resp. SFmode cases.
25989 2018-03-15 Tamar Christina <tamar.christina@arm.com>
25992 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
25993 instead of GET_MODE_SIZE when comparing Units.
25995 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
25998 * varasm.c (hash_section): Return an unchangeble hash value
25999 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
26000 Return !aarch64_can_use_per_function_literal_pools_p ().
26002 2018-03-15 Jakub Jelinek <jakub@redhat.com>
26005 * optabs.c (emit_conditional_move): Pass address of cmode's copy
26006 rather than address of cmode as last argument to prepare_cmp_insn.
26008 2018-03-15 Julia Koval <julia.koval@intel.com>
26010 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
26011 F_AVX512VNNI, F_AVX512BITALG): New.
26013 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
26016 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
26017 insn for floating-point loads and stores.
26019 2018-03-14 Carl Love <cel@us.ibm.com>
26021 * config/rs6000/rs6000-c.c: Add macro definitions for
26022 ALTIVEC_BUILTIN_VEC_PERMXOR.
26023 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
26024 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
26025 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
26026 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
26028 * config/doc/extend.texi: Add prototypes for vec_permxor.
26030 2018-03-14 David Malcolm <dmalcolm@redhat.com>
26033 * diagnostic-show-locus.c (class layout_point): Convert m_line
26034 from int to linenum_type.
26035 (line_span::comparator): Use linenum "compare" function when
26036 comparing line numbers.
26037 (test_line_span): New function.
26038 (layout_range::contains_point): Convert param "row" from int to
26040 (layout_range::intersects_line_p): Likewise.
26041 (layout::will_show_line_p): Likewise.
26042 (layout::print_source_line): Likewise.
26043 (layout::should_print_annotation_line_p): Likewise.
26044 (layout::print_annotation_line): Likewise.
26045 (layout::print_leading_fixits): Likewise.
26046 (layout::annotation_line_showed_range_p): Likewise.
26047 (struct line_corrections): Likewise for field m_row.
26048 (line_corrections::line_corrections): Likewise for param "row".
26049 (layout::print_trailing_fixits): Likewise.
26050 (layout::get_state_at_point): Likewise.
26051 (layout::get_x_bound_for_row): Likewise.
26052 (layout::print_line): Likewise.
26053 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
26054 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
26055 * input.c (selftest::test_linenum_comparisons): New function.
26056 (selftest::input_c_tests): Call it.
26057 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
26058 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
26059 * selftest.h (ASSERT_GT): New macro.
26060 (ASSERT_GT_AT): New macro.
26061 (ASSERT_LT): New macro.
26062 (ASSERT_LT_AT): New macro.
26064 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
26066 PR rtl-optimization/84780
26067 * combine.c (distribute_links): Don't make a link based on pc_rtx.
26069 2018-03-14 Martin Liska <mliska@suse.cz>
26071 * tree.c (record_node_allocation_statistics): Use
26072 get_stats_node_kind.
26073 (get_stats_node_kind): New function extracted from
26074 record_node_allocation_statistics.
26075 (free_node): Use get_stats_node_kind.
26077 2018-03-14 Richard Biener <rguenther@suse.de>
26079 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
26080 that the value-set of ANTIC_IN doesn't grow.
26083 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
26085 (BB_VISITED_WITH_VISITED_SUCCS): New define.
26086 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
26088 2018-03-14 Julia Koval <julia.koval@intel.com>
26090 * config.gcc (icelake-client, icelake-server): New.
26092 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
26093 (initial_ix86_arch_features): Ditto.
26094 (PTA_SKYLAKE): Add SGX.
26095 (PTA_ICELAKE): Remove.
26096 (PTA_ICELAKE_CLIENT): New.
26097 (PTA_ICELAKE_SERVER): New.
26098 (ix86_option_override_internal): Split up icelake on icelake client and
26100 (get_builtin_code_for_version): Ditto.
26101 (fold_builtin_cpu): Ditto.
26102 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
26103 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
26104 * config/i386/i386.h (processor_type): Ditto.
26105 * doc/invoke.texi: Ditto.
26107 2018-03-14 Jakub Jelinek <jakub@redhat.com>
26110 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
26111 INTEGER_CST offset, add it together with bitpos / 8 and
26112 sign extend based on POINTER_SIZE.
26116 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26119 * config/i386/constraints.md (Yc): New register constraint.
26120 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26121 Use Yc constraint for alternative 2 of operand 0. Remove
26122 preferred_for_speed attribute.
26124 2018-03-14 Richard Biener <rguenther@suse.de>
26126 PR tree-optimization/84830
26127 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
26128 with the old one to avoid oscillations.
26130 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
26133 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
26135 (assign_by_spills): Return a flag of reload assignment failure.
26136 Do not process the reload assignment failures. Do not spill other
26137 reload pseudos if they has the same reg class. Update n if
26139 (lra_assign): Add a return arg. Set up from the result of
26140 assign_by_spills call.
26141 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
26142 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
26143 usage_insns if it is not NULL.
26144 (spill_hard_reg_in_range): New function.
26145 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
26146 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
26147 function prototypes.
26148 (lra_assign): Change prototype.
26149 * lra.c (lra): Add code to deal with fails by splitting hard reg
26152 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
26154 * config/riscv/riscv.opt (mrelax): New option.
26155 * config/riscv/riscv.c (riscv_file_start): Emit ".option
26156 "norelax" when riscv_mrelax is disabled.
26157 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
26159 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26162 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
26163 reassociation for int modes.
26165 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
26167 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
26168 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
26170 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
26171 * config/aarch64/aarch64-sve.md
26172 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
26173 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
26174 (*extend<mode><Vwide>2): Rename to...
26175 (aarch64_sve_extend<mode><Vwide>2): ...this.
26176 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
26177 renaming the old pattern to...
26178 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
26180 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
26181 define_expand, renaming the old pattern to...
26182 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
26183 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
26184 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
26185 account when deciding which SVE instruction the optab should use.
26186 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
26188 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
26190 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
26191 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
26192 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
26193 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
26194 (tlsdesc_small_<mode>): Turn a define_expand and use
26195 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
26196 (tlsdesc_small_advsimd_<mode>): ...this.
26197 (tlsdesc_small_sve_<mode>): New pattern.
26199 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
26201 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
26202 (UNSPEC_UMUL_HIGHPART): New constants.
26203 (MUL_HIGHPART): New int iteraor.
26204 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
26205 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
26207 (*<su>mul<mode>3_highpart): New define_insn.
26209 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
26212 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
26215 2018-03-13 Martin Liska <mliska@suse.cz>
26218 * (sem_item_optimizer::sem_item_optimizer): Initialize new
26220 (sem_item_optimizer::~sem_item_optimizer): Release it.
26221 (sem_item_optimizer::merge_classes): Register variable aliases.
26222 (sem_item_optimizer::fixup_pt_set): New function.
26223 (sem_item_optimizer::fixup_points_to_sets): Likewise.
26224 * ipa-icf.h: Declare new variables and functions.
26226 2018-03-13 Jakub Jelinek <jakub@redhat.com>
26228 PR middle-end/84834
26229 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
26230 integer_pow2p@2 and test integer_pow2p in condition.
26231 (A < 0 ? C : 0): Similarly for @1.
26233 PR middle-end/84831
26234 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
26235 characters starting at p contain '\0' character, don't look beyond
26239 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
26240 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
26243 * reg-stack.c (change_stack): Change update_end var from int to
26244 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
26245 also call set_block_for_insn on the newly added insns and rescan.
26248 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
26249 on the last operand.
26252 * tree.c (stabilize_reference_1): Return save_expr (e) for
26253 STATEMENT_LIST even if it doesn't have side-effects.
26255 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
26257 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
26259 2018-03-12 Renlin Li <renlin.li@arm.com>
26261 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
26262 aarch64_output_scalar_simd_mov_immediate.
26264 2018-03-12 Martin Sebor <msebor@redhat.com>
26266 PR tree-optimization/83456
26267 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
26268 for perfectly overlapping calls to memcpy.
26269 (gimple_fold_builtin_memory_chk): Same.
26270 (gimple_fold_builtin_strcpy): Handle no-warning.
26271 (gimple_fold_builtin_stxcpy_chk): Same.
26272 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
26274 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
26276 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
26277 parameter. Use it for SFmode.
26278 (rs6000_function_arg_advance_1): Adjust.
26279 (rs6000_function_arg): Adjust.
26280 (rs6000_gimplify_va_arg): Pass false for that new parameter.
26282 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
26284 PR rtl-optimization/84169
26285 PR rtl-optimization/84780
26286 * combine.c (can_combine_p): Check for a 2-insn combination whether
26287 the destination register is used between the two insns, too.
26289 2018-03-12 Richard Biener <rguenther@suse.de>
26291 PR tree-optimization/84803
26292 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
26293 for refs DR analysis didn't process.
26295 2018-03-12 Richard Biener <rguenther@suse.de>
26297 PR tree-optimization/84777
26298 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
26299 force-vectorize loops ignore whether we are optimizing for size.
26301 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
26303 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
26304 (TARGET_MD_ASM_ADJUST): Define.
26306 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
26307 Kito Cheng <kito.cheng@gmail.com>
26308 Chung-Ju Wu <jasonwucj@gmail.com>
26310 * config/nds32/nds32.c (nds32_compute_stack_frame,
26311 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
26312 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
26313 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
26314 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
26315 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
26316 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
26317 * config/nds32/nds32.md (prologue, epilogue): Use macro
26318 NDS32_V3PUSH_AVAILABLE_P to do checking.
26320 2018-03-11 Jakub Jelinek <jakub@redhat.com>
26323 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
26324 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
26325 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
26326 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
26327 addition of most attributes on !orig_type_die or the attribute not
26328 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
26330 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
26331 Chung-Ju Wu <jasonwucj@gmail.com>
26333 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
26334 __NDS32_VH__ macro.
26335 * config/nds32/nds32.opt (mvh): New option.
26337 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
26338 Chung-Ju Wu <jasonwucj@gmail.com>
26340 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
26342 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
26343 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
26346 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
26347 Chung-Ju Wu <jasonwucj@gmail.com>
26349 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
26351 * config/nds32/nds32-multiple.md (strlensi): New pattern.
26352 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
26354 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
26355 Kito Cheng <kito.cheng@gmail.com>
26356 Chung-Ju Wu <jasonwucj@gmail.com>
26358 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
26359 UNSPEC_FFMISM and UNSPEC_FLMISM.
26360 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
26361 for ffb, ffmism and flmism.
26362 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
26363 (unspec_ffmism): Ditto.
26364 (unspec_flmism): Ditto.
26365 (nds32_expand_builtin_impl): Check if string extension is available.
26366 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
26367 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
26369 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
26372 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
26375 * lra-assigns.c (assign_by_spills): Return a flag of reload
26376 assignment failure. Do not process the reload assignment
26377 failures. Do not spill other reload pseudos if they has the same
26379 (lra_assign): Add a return arg. Set up from the result of
26380 assign_by_spills call.
26381 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
26382 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
26383 usage_insns if it is not NULL.
26384 (spill_hard_reg_in_range): New function.
26385 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
26386 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
26387 function prototypes.
26388 (lra_assign): Change prototype.
26389 * lra.c (lra): Add code to deal with fails by splitting hard reg
26392 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
26395 * config/i386/i386.opt: Replace Enforcment with Enforcement.
26397 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
26400 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
26401 (dw_val_node): Add val_symbolic_view.
26402 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
26403 (symview_upper_bound): New.
26404 (new_line_info_table): Initialize symviews_since_reset.
26405 (dwarf2out_source_line): Count symviews_since_reset and set
26406 symview_upper_bound.
26407 (dw_val_equal_p): Handle symview.
26408 (add_AT_symview): New.
26409 (print_dw_val): Handle symview.
26410 (attr_checksum, attr_checksum_ordered): Likewise.
26411 (same_dw_val_p, size_of_die): Likewise.
26412 (value_format, output_die): Likewise.
26413 (add_high_low_attributes): Use add_AT_symview for entry_view.
26414 (dwarf2out_finish): Reset symview_upper_bound, clear
26417 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
26420 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
26421 Add strict argument and use it.
26422 (rs6000_split_multireg_move): Update for new strict argument.
26423 (mem_operand_gpr): Disallow all non-offsettable addresses.
26424 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
26426 2018-03-09 Jakub Jelinek <jakub@redhat.com>
26429 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
26430 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
26431 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
26434 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
26435 decl, use remap_type if we want to use the type.
26437 2018-03-09 Martin Sebor <msebor@redhat.com>
26439 PR tree-optimization/84526
26440 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
26442 (builtin_access::generic_overlap): Be prepared to handle non-array
26445 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
26447 PR rtl-optimization/84682
26448 * lra-constraints.c (process_address_1): Check is_address flag
26449 for address constraints.
26450 (process_alt_operands): Likewise.
26451 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
26452 preprocess_constraints.
26453 * recog.h (preprocess_constraints): Add oploc parameter.
26455 * recog.c (preprocess_constraints): Test address_operand for
26456 CT_ADDRESS constraints.
26458 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
26461 * lra-assigns.c (assign_by_spills): Return a flag of reload
26462 assignment failure. Do not process the reload assignment
26463 failures. Do not spill other reload pseudos if they has the same
26465 (lra_assign): Add a return arg. Set up from the result of
26466 assign_by_spills call.
26467 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
26468 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
26469 usage_insns if it is not NULL.
26470 (spill_hard_reg_in_range): New function.
26471 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
26472 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
26473 function prototypes.
26474 (lra_assign): Change prototype.
26475 * lra.c (lra): Add code to deal with fails by splitting hard reg
26478 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26481 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
26482 Accept complain bool parameter. Only emit errors if it is true.
26483 (arm_parse_cpu_option_name): Likewise.
26484 (arm_target_thumb_only): Adjust callers of the above.
26485 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
26486 prototype to take a default true bool parameter.
26487 (arm_parse_arch_option_name): Likewise.
26489 2018-03-09 David Malcolm <dmalcolm@redhat.com>
26490 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
26494 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
26495 * configure: Regenerate.
26496 * configure.ac ("linker --version-script option"): New.
26497 ("linker soname option"): New.
26499 2018-03-09 Richard Biener <rguenther@suse.de>
26501 PR tree-optimization/84775
26502 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
26503 immediate uses of predicate stmts and mark them modified.
26506 PR tree-optimization/84178
26507 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
26509 (version_loop_for_if_conversion): Delay update_ssa call.
26510 (tree_if_conversion): Delay update_ssa until after predicate
26513 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
26516 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
26517 when the function accesses prior frames.
26519 2018-03-08 Jakub Jelinek <jakub@redhat.com>
26522 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
26523 gen_llsym, otherwise call maybe_gen_llsym.
26525 PR inline-asm/84742
26526 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
26527 has ',' character inside of it.
26529 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26532 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
26533 as clobbering CC_REGNUM.
26535 2018-03-08 Richard Biener <rguenther@suse.de>
26537 PR middle-end/84552
26538 * tree-scalar-evolution.c: Include tree-into-ssa.h.
26539 (follow_copies_to_constant): Do not follow SSA names registered
26542 2018-03-08 Richard Biener <rguenther@suse.de>
26544 PR tree-optimization/84178
26545 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
26547 (version_loop_for_if_conversion): Delay update_ssa call.
26548 (tree_if_conversion): Delay update_ssa until after predicate
26551 2018-03-08 David Malcolm <dmalcolm@redhat.com>
26553 PR tree-optimization/84178
26554 * tree-if-conv.c (release_bb_predicate): Remove the
26555 the assertion that the stmts have NULL use_ops.
26556 Discard the statements, asserting that they haven't
26557 yet been added to a BB.
26559 2018-03-08 Richard Biener <rguenther@suse.de>
26561 PR tree-optimization/84746
26562 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
26563 (phi_translate): Pass in destination ANTIC_OUT set.
26564 (phi_translate_1): Likewise. For a simplified result lookup
26565 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
26566 (phi_translate_set): Adjust.
26567 (do_pre_regular_insertion): Likewise.
26568 (do_pre_partial_partial_insertion): Likewise.
26570 2018-03-08 Martin Liska <mliska@suse.cz>
26572 PR gcov-profile/84735
26573 * doc/gcov.texi: Document usage of profile files.
26574 * gcov-io.h: Document changes in the format.
26576 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
26580 * dwarf2out.c (struct dw_line_info_table): Update comments for
26582 (FORCE_RESET_NEXT_VIEW): New.
26583 (FORCE_RESETTING_VIEW_P): New.
26584 (RESETTING_VIEW_P): Check for -1 too.
26585 (ZERO_VIEW_P): Likewise.
26586 (new_line_info_table): Force-reset next view.
26587 (dwarf2out_begin_function): Likewise.
26588 (dwarf2out_source_line): Simplify zero_view_p initialization.
26589 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
26590 view directly. Omit view when omitting .loc at line 0.
26592 2018-03-08 Jakub Jelinek <jakub@redhat.com>
26594 PR tree-optimization/84740
26595 * tree-switch-conversion.c (process_switch): Call build_constructors
26596 only if info.phi_count is non-zero.
26598 PR tree-optimization/84739
26599 * tree-tailcall.c (find_tail_calls): Check call arguments against
26600 DECL_ARGUMENTS (current_function_decl) rather than
26601 DECL_ARGUMENTS (func) when checking for tail recursion.
26603 2018-03-07 Jakub Jelinek <jakub@redhat.com>
26605 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
26606 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
26607 Volker Reichelt's entry and add entries for people that perform
26608 GCC fuzzy testing and report numerous bugs.
26610 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
26613 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
26614 readonly data in sdata, if that is disabled.
26615 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
26616 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
26617 -mreadonly-in-sdata option.
26619 2018-03-07 Martin Sebor <msebor@redhat.com>
26621 PR tree-optimization/84468
26622 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
26623 basic block when looking for nul assignment.
26625 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
26628 * except.h (output_function_exception_table): Adjust prototype.
26629 * except.c (output_function_exception_table): Remove FNNAME parameter
26630 and add SECTION parameter. Ouput one part of the table at a time.
26631 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
26632 the first part of the exception table and emit unwind directives.
26633 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
26634 (i386_pe_seh_cold_init): Likewise.
26635 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
26636 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
26637 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
26638 (ix86_output_call_insn): Emit a nop in one more case for SEH.
26639 * config/i386/winnt.c: Include except.h.
26640 (struct seh_frame_state): Add reg_offset, after_prologue and
26641 in_cold_section fields.
26642 (i386_pe_seh_end_prologue): Set seh->after_prologue.
26643 (i386_pe_seh_cold_init): New function.
26644 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
26645 to seh->in_cold_section.
26646 (seh_emit_push): Record the offset of the push.
26647 (seh_emit_save): Record the offet of the save.
26648 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
26649 Test seh->after_prologue to disregard the epilogue.
26650 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
26651 (i386_pe_end_cold_function): New function.
26653 2018-03-07 Jakub Jelinek <jakub@redhat.com>
26656 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
26657 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
26660 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
26662 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
26663 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
26665 PR middle-end/84723
26666 * multiple_target.c: Include tree-inline.h and intl.h.
26667 (expand_target_clones): Diagnose and fail if node->definition and
26668 !tree_versionable_function_p (node->decl).
26670 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
26672 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
26674 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
26675 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
26676 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
26678 2018-03-06 Jakub Jelinek <jakub@redhat.com>
26681 * combine.c (try_combine): Use reg_or_subregno instead of handling
26682 just paradoxical SUBREGs and REGs.
26684 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
26686 * config/arc/arc.c (arc_finalize_pic): Remove function.
26687 (arc_must_save_register): We use single base PIC register, remove
26688 checks to save/restore the PIC register.
26689 (arc_expand_prologue): Likewise.
26690 * config/arc/arc-protos.h (arc_set_default_type_attributes):
26692 (arc_verify_short): Likewise.
26693 (arc_attr_type): Likewise.
26694 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
26695 (walk_stores): Likewise.
26696 (arc_address_cost): Make it static.
26697 (arc_verify_short): Likewise.
26698 (branch_dest): Likewise.
26699 (arc_attr_type): Likewise.
26700 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
26701 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
26702 (arc_final_prescan_insn): Remove inserting the nops due to
26703 hardware hazards. It is done in reorg step.
26704 (insn_length_variant_t): Remove.
26705 (insn_length_parameters_t): Likewise.
26706 (arc_insn_length_parameters): Likewise.
26707 (arc_get_insn_variants): Likewise.
26708 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
26710 2018-03-06 Jakub Jelinek <jakub@redhat.com>
26712 PR inline-asm/84683
26713 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
26716 PR tree-optimization/84687
26717 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
26719 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
26721 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26723 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
26724 Rename to ppc_speculation_barrier.
26725 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
26726 __builtin_ppc_speculation_barrier.
26728 2018-03-05 Jakub Jelinek <jakub@redhat.com>
26731 * combine.c (combine_simplify_rtx): Don't try to simplify if
26732 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
26735 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
26737 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
26738 to 32 bytes when compiling for POWER9.
26740 2018-03-05 Jakub Jelinek <jakub@redhat.com>
26743 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
26744 regparm >= 3 with no arg reg available also for calls with
26745 flag_force_indirect_call. Pass decl to ix86_function_regparm.
26748 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
26750 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
26752 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
26755 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
26757 2018-03-05 Richard Biener <rguenther@suse.de>
26759 PR tree-optimization/84486
26760 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
26761 When inserting a __builtin_assume_aligned call set the LHS
26762 SSA name alignment info accordingly.
26764 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
26766 PR tree-optimization/84114
26767 * config/aarch64/aarch64.c (aarch64_reassociation_width)
26768 Avoid reassociation of FLOAT_MODE addition.
26770 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
26772 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
26773 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
26774 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
26775 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
26776 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
26777 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
26778 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
26780 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
26781 __builtin_ia32_wbinvd): New builtins.
26782 (SPECIAL_ARGS2): New.
26783 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
26784 (SPECIAL_ARGS2): New.
26785 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
26786 (ix86_valid_target_attribute_inner_p): Ditto.
26787 (ix86_init_mmx_sse_builtins): Add special_args2.
26788 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
26789 TARGET_WBNOINVD_P): New.
26790 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
26791 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
26792 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
26793 * config/i386/immintrin.h (_wbinvd): New intrinsic.
26794 * config/i386/pconfigintrin.h: New file.
26795 * config/i386/wbnoinvdintrin.h: Ditto.
26796 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
26798 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
26800 2018-03-05 Richard Biener <rguenther@suse.de>
26802 PR tree-optimization/84670
26803 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
26805 (BB_VISITED_WITH_VISITED_SUCCS): New define.
26806 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
26807 (compute_antic_aux): Only assert the number of values in ANTIC_IN
26808 doesn't grow if all successors (recursively) were visited at least
26811 2018-03-05 Richard Biener <rguenther@suse.de>
26813 PR tree-optimization/84650
26814 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
26815 if executed in the loop pipeline.
26817 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
26819 * doc/configfiles.texi (Configuration Files): Move info about
26820 conditionalizing $target-protos.h to...
26821 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
26822 differs from $target-protos.h.
26824 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
26825 Chung-Ju Wu <jasonwucj@gmail.com>
26827 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
26828 * config/nds32/nds32-multiple.md (setmemsi): Define.
26829 * config/nds32/nds32-memory-manipulation.c
26830 (nds32_gen_dup_4_byte_to_word_value): New.
26831 (emit_setmem_word_loop): New.
26832 (emit_setmem_byte_loop): New.
26833 (nds32_expand_setmem_loop): New.
26834 (nds32_expand_setmem_loop_v3m): New.
26835 (nds32_expand_setmem_unroll): New.
26836 (nds32_expand_setmem): New.
26838 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
26839 Chung-Ju Wu <jasonwucj@gmail.com>
26841 * config/nds32/nds32-memory-manipulation.c
26842 (nds32_emit_load_store): New.
26843 (nds32_emit_post_inc_load_store): New.
26844 (nds32_emit_mem_move): New.
26845 (nds32_emit_mem_move_block): New.
26846 (nds32_expand_movmemsi_loop_unknown_size): New.
26847 (nds32_expand_movmemsi_loop_known_size): New.
26848 (nds32_expand_movmemsi_loop): New.
26849 (nds32_expand_movmemsi_unroll): New.
26850 (nds32_expand_movmemqi): Rename ...
26851 (nds32_expand_movmemsi): ... to this.
26852 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
26853 (movmemsi): ... to this.
26854 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
26855 (nds32_expand_movmemsi): ... to this.
26857 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
26858 Monk Chiang <sh.chiang04@gmail.com>
26859 Chung-Ju Wu <jasonwucj@gmail.com>
26861 * config/nds32/nds32-protos.h
26862 (nds32_expand_load_multiple): New arguments.
26863 (nds32_expand_store_multiple): Ditto.
26864 (nds32_valid_multiple_load_store): Rename ...
26865 (nds32_valid_multiple_load_store_p): ... to this.
26866 * config/nds32/nds32-memory-manipulation.c
26867 (nds32_expand_load_multiple): Refine implementation.
26868 (nds32_expand_store_multiple): Ditto.
26869 * config/nds32/nds32-multiple.md
26870 (load_multiple): Update nds32_expand_load_multiple interface.
26871 (store_multiple): Update nds32_expand_store_multiple interface.
26872 * config/nds32/nds32-predicates.c
26873 (nds32_valid_multiple_load_store): Rename ...
26874 (nds32_valid_multiple_load_store_p): ... to this and refine
26876 * config/nds32/predicates.md
26877 (nds32_load_multiple_and_update_address_operation): New predicate.
26878 (nds32_store_multiple_and_update_address_operation): New predicate.
26880 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
26881 Chung-Ju Wu <jasonwucj@gmail.com>
26883 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
26884 (combo): New attribute.
26885 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
26887 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
26889 * config/nds32/nds32.opt: Change -mcmodel= default value.
26891 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
26892 Monk Chiang <sh.chiang04@gmail.com>
26893 Chung-Ju Wu <jasonwucj@gmail.com>
26895 * config/nds32/constants.md (unspec_element): New enum.
26896 * config/nds32/constraints.md (Umw): New constraint.
26897 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
26898 * config/nds32/nds32-intrinsic.md: Likewise.
26899 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
26900 (nds32_valid_smw_lwm_base_p): New.
26901 (nds32_output_smw_single_word): New.
26902 (nds32_output_lmw_single_word): New.
26903 (nds32_expand_unaligned_load): New.
26904 (nds32_expand_unaligned_store): New.
26905 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
26906 (nds32_output_smw_single_word): Declare.
26907 (nds32_output_lmw_single_word): Declare.
26908 (nds32_expand_unaligned_load): Declare.
26909 (nds32_expand_unaligned_store): Declare.
26910 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
26911 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
26912 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
26913 NDS32_BUILTIN_UASTORE_DW.
26914 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
26917 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
26918 Kito Cheng <kito.cheng@gmail.com>
26919 Chung-Ju Wu <jasonwucj@gmail.com>
26921 * config/nds32/nds32-intrinsic.c
26922 (nds32_expand_builtin_null_ftype_reg): Delete.
26923 (nds32_expand_builtin_reg_ftype_imm): Ditto.
26924 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
26925 (nds32_read_argument): New.
26926 (nds32_legitimize_target): Ditto.
26927 (nds32_legitimize_argument): Ditto.
26928 (nds32_check_constant_argument): Ditto.
26929 (nds32_expand_unop_builtin): Ditto.
26930 (nds32_expand_unopimm_builtin): Ditto.
26931 (nds32_expand_binop_builtin): Ditto.
26932 (nds32_builtin_decl_impl): Ditto.
26933 (builtin_description): Ditto.
26934 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
26935 (nds32_init_builtins_impl): Ditto.
26936 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
26937 (nds32_builtin_decl): New.
26938 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
26939 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
26941 2018-03-02 Jeff Law <law@redhat.com>
26943 * reorg.c (stop_search_p): Handle DEBUG_INSN.
26944 (redundant_insn, fill_simple_delay_slots): Likewise.
26945 (fill_slots_from_thread): Likewise.
26946 * resource.c (mark_referenced_resources): Likewise.
26947 (mark_set_resources, find_dead_or_set_registers): Likewise.
26949 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26951 * substring-locations.h (format_warning_va): Formatting fix for
26952 ATTRIBUTE_GCC_DIAG.
26953 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
26955 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
26956 * substring-locations.c: Include intl.h.
26957 (format_warning_va): Turned into small wrapper around
26958 format_warning_n_va, renamed to ...
26959 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
26960 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
26962 (format_warning_at_substring_n): New function.
26963 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
26964 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
26965 format_warning_at_substring with just a shorter name instead of
26966 const function pointer.
26967 (fmtwarn_n): New function.
26968 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
26969 appropriate, get rid of all the fmtstr temporaries, move conditionals
26970 with G_() wrapped string literals directly into fmtwarn arguments,
26971 cast dir.len to (int), formatting fixes.
26973 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
26975 * doc/invoke.texi: Remove "Cilk Plus" references.
26977 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26978 Richard Biener <rguenther@suse.de>
26981 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
26982 for error or warning attributes if CALL_FROM_THUNK_P is set.
26985 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26988 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
26989 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
26992 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
26993 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
26995 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
26996 instead of -1U in last predictors element's probability member.
26998 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
27001 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
27002 arguments if they are comparable.
27004 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
27006 PR tree-optimization/84634
27007 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
27008 masks and masked_loop_p with a single loop_masks, making sure it's
27009 null for bb vectorization.
27011 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
27013 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
27014 (vect_analyze_data_ref_access): Use loop->safe_len rather than
27015 loop->force_vectorize to check whether there is no alias.
27017 2018-03-02 Jakub Jelinek <jakub@redhat.com>
27020 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
27022 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
27024 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
27025 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
27026 instead of a loop around prev_real_insn.
27027 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
27030 PR inline-asm/84625
27031 * config/i386/i386.c (ix86_print_operand): Use conditional
27032 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
27035 2018-03-02 Richard Biener <rguenther@suse.de>
27037 PR tree-optimization/84427
27038 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
27039 (bitmap_set_subtract_values): Rewrite to handle multiple
27042 (prune_clobbered_mems): Likewise.
27043 (phi_translate): Take edge instead of pred/phiblock.
27044 (phi_translate_1): Likewise.
27045 (phi_translate_set): Likewise. Insert all translated
27046 exprs for a value into the set, keeping possibly multiple
27047 expressions per value.
27048 (compute_antic_aux): Adjust for phi_translate changes.
27049 When intersecting union the expressions and prune those
27050 not in the final value set, keeping possibly multiple
27051 expressions per value. Do not use value-insertion
27052 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
27053 all expressions. Add verification that the value-sets
27054 only shrink during iteration.
27055 (compute_partial_antic_aux): Adjust for the phi_translate changes.
27056 (do_pre_regular_insertion): Likewise.
27057 (do_pre_partial_partial_insertion): Likewise.
27059 2018-03-02 Richard Biener <rguenther@suse.de>
27062 * config/darwin.c (saved_debug_info_level): New static global.
27063 (darwin_asm_lto_start): Disable debug info generation for LTO out.
27064 (darwin_asm_lto_end): Restore debug info generation settings.
27066 2018-03-01 Martin Liska <mliska@suse.cz>
27069 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
27070 volatile arguments.
27072 2018-03-01 Richard Biener <rguenther@suse.de>
27075 * dwarf2out.c (gen_variable_die): Properly handle late VLA
27076 type annotation with LTO when debug was disabled at compile-time.
27078 2018-03-01 Matthew Fortune <mfortune@gmail.com>
27080 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
27082 (mips_final_postscan_insn): Likewise.
27084 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
27086 PR rtl-optimization/84528
27087 * alias.c (init_alias_target): Add commentary.
27088 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
27089 a unique base value if the frame pointer is not eliminated
27090 to the stack pointer.
27092 2018-03-01 Tom de Vries <tom@codesourcery.com>
27094 PR rtl-optimization/83327
27095 * lra-int.h (hard_regs_spilled_into): Declare.
27096 * lra.c (hard_regs_spilled_into): Define.
27097 (init_reg_info): Init hard_regs_spilled_into.
27098 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
27099 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
27100 (process_bb_lives): Handle hard_regs_spilled_into.
27101 (lra_create_live_ranges_1): Before doing liveness propagation, clear
27102 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
27104 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
27106 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
27107 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
27108 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
27109 * config/rs6000/aix72.h: New file.
27111 2018-02-28 Jakub Jelinek <jakub@redhat.com>
27113 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
27114 instead of warning_at with conditional singular and plural messages
27118 * stor-layout.c (update_alignment_for_field): For
27119 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
27120 && !DECL_PACKED (field), do the alignment update, just use
27121 only desired_align instead of MAX (type_align, desired_align)
27123 (place_field): Don't do known_align < desired_align handling
27124 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
27125 is non-NULL, instead do it after rli->prev_field handling and
27126 only if not within a bitfield word. For DECL_PACKED (field)
27127 use type_align of BITS_PER_UNIT.
27129 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
27131 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
27132 superfluous parentheses and trailing spaces.
27134 2018-02-28 Richard Biener <rguenther@suse.de>
27136 PR tree-optimization/84584
27137 * graphite-scop-detection.c (scop_detection::add_scop): Discard
27138 SCoPs with fake exit edge.
27140 2018-02-28 Martin Liska <mliska@suse.cz>
27143 * timevar.c (timer::print): Fix format to properly print 100%
27146 2018-02-28 Richard Biener <rguenther@suse.de>
27148 PR middle-end/84607
27149 * genmatch.c (capture_info::walk_match): Do not mark
27150 captured expressions without operands as expr_p given
27151 they act more like predicates and should be subject to
27152 "lost tail" side-effect preserving.
27154 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
27156 PR rtl-optimization/81611
27157 * auto-inc-dec.c (attempt_change): Move dead note from
27158 mem_insn if it's the next use of regno
27159 (find_address): Take address use of reg holding
27160 non-incremented value. Add parm to limit search to the named
27162 (merge_in_block): Attempt to use a mem insn that is the next
27163 use of the original regno.
27165 2018-02-27 Martin Sebor <msebor@redhat.com>
27168 * doc/invoke.texi (-Wmissing-attributes): New option.
27169 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
27171 2018-02-27 Martin Sebor <msebor@redhat.com>
27173 PR translation/84207
27174 * diagnostic-core.h (warning_n, error_n, inform_n): Change
27175 n argument to unsigned HOST_WIDE_INT.
27176 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
27177 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
27178 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
27179 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
27181 2018-02-27 Richard Biener <rguenther@suse.de>
27183 PR tree-optimization/84512
27184 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
27185 Do not use the estimate returned from record_stmt_cost for
27186 the scalar iteration cost but sum properly using add_stmt_cost.
27188 2018-02-27 Richard Biener <rguenther@suse.de>
27190 PR tree-optimization/84466
27191 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
27192 Adjust last change to less strictly validate use operands.
27194 2018-02-27 Martin Liska <mliska@suse.cz>
27196 PR gcov-profile/84548
27197 * gcov.c (process_file): Allow partial overlap and consider it
27198 also as group functions.
27199 (output_lines): Properly calculate range of lines for a group.
27201 2018-02-27 Martin Liska <mliska@suse.cz>
27203 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
27204 'ggc' suffixes. Change first column width.
27205 (timer::print): Fix formatting of the column.
27207 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
27209 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
27210 preserve inline entry blocks for the sake of debug inline
27211 entry point markers alone.
27212 (remove_unused_locals): Suggest in comments a better place to
27213 force the preservation of inline entry blocks that are
27214 otherwise unused, but do not preserve them.
27216 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
27218 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
27220 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
27223 * config/i386/constraints.md (Bs): Replace
27224 ix86_indirect_branch_register with
27225 TARGET_INDIRECT_BRANCH_REGISTER.
27227 * config/i386/i386.md (indirect_jump): Likewise.
27228 (tablejump): Likewise.
27229 (*sibcall_memory): Likewise.
27230 (*sibcall_value_memory): Likewise.
27231 Peepholes of indirect call and jump via memory: Likewise.
27232 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
27233 (*sibcall_value_GOT_32): Likewise.
27234 * config/i386/predicates.md (indirect_branch_operand): Likewise.
27235 (GOT_memory_operand): Likewise.
27236 (call_insn_operand): Likewise.
27237 (sibcall_insn_operand): Likewise.
27238 (GOT32_symbol_operand): Likewise.
27239 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
27241 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
27243 PR rtl-optimization/83496
27244 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
27245 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
27246 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
27247 redundant insn, if any.
27248 (relax_delay_slots): Likewise.
27249 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
27251 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
27253 PR tree-optimization/83965
27254 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
27255 that grouped statements are part of a reduction chain. Return
27256 true if the statement is not marked as a reduction itself but
27257 is part of a group.
27258 (vect_recog_dot_prod_pattern): Don't check whether the statement
27259 is part of a group here.
27260 (vect_recog_sad_pattern): Likewise.
27261 (vect_recog_widen_sum_pattern): Likewise.
27263 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
27266 * final.c (rest_of_clean_state): Also look for calls inside sequences.
27268 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
27271 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
27273 (ix86_output_indirect_function_return): New prototype.
27274 (ix86_split_simple_return_pop_internal): Likewise.
27275 * config/i386/i386.c (indirect_return_via_cx): New.
27276 (indirect_return_via_cx_bnd): Likewise.
27277 (indirect_thunk_name): Handle return va CX_REG.
27278 (output_indirect_thunk_function): Create alias for
27279 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
27280 (ix86_output_indirect_jmp): Remove the bool argument.
27281 (ix86_output_indirect_function_return): New function.
27282 (ix86_split_simple_return_pop_internal): Likewise.
27283 * config/i386/i386.md (*indirect_jump): Don't pass false
27284 to ix86_output_indirect_jmp.
27285 (*tablejump_1): Likewise.
27286 (simple_return_pop_internal): Change it to define_insn_and_split.
27287 Call ix86_split_simple_return_pop_internal to split it for
27288 -mfunction-return=.
27289 (simple_return_indirect_internal): Call
27290 ix86_output_indirect_function_return instead of
27291 ix86_output_indirect_jmp.
27293 2018-02-26 Jakub Jelinek <jakub@redhat.com>
27296 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
27297 memset and value initialization afterwards.
27299 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
27301 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
27303 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27306 * common/config/aarch64/aarch64-common.c
27307 (aarch_option_optimization_table[]): Switch
27308 off fomit-frame-pointer
27310 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
27311 Chung-Ju Wu <jasonwucj@gmail.com>
27313 * config/nds32/nds32-multiple.md (load_multiple): Disallow
27315 (store_multiple): Ditto.
27317 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
27319 * config.gcc: Add --with-cpu support for nds32 target.
27320 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
27321 * config/nds32/nds32.opt: Add -mcpu= option.
27323 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
27325 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
27326 isel=yes): Warn for these deprecated options.
27328 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
27330 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
27331 ISA_2_5_MASKS_EMBEDDED.
27333 2018-02-23 Jakub Jelinek <jakub@redhat.com>
27335 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
27336 p->max as pointers rather than using iterative_hash_expr.
27338 2018-02-23 Carl Love <cel@us.ibm.com>
27340 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
27341 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
27343 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
27344 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
27345 P8V_BUILTIN_VEC_VUNSIGNED2.
27347 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
27350 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
27351 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
27353 * lra-constraints.c (curr_insn_transform): Set up
27354 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
27356 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
27357 * lra-eliminations.c (spill_pseudos): Ditto.
27358 (process_insn_for_elimination): Ditto.
27359 * lra-lives.c (reg_early_clobber_p): Use the new macros.
27360 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
27361 LRA_NON_CLOBBERED_ALT.
27363 2018-02-22 Martin Sebor <msebor@redhat.com>
27365 PR tree-optimization/84480
27366 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
27367 to maybe_diag_stxncpy_trunc. Call it.
27368 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
27369 from gimple_fold_builtin_strcpy. Print inlining stack.
27370 (handle_builtin_stxncpy): Print inlining stack.
27371 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
27373 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
27376 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
27377 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
27378 and -fcheck-pointer-bounds are used together.
27379 (indirect_thunk_prefix): New enum.
27380 (indirect_thunk_need_prefix): New function.
27381 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
27382 "_nt" instead of "_bnd" for NOTRACK prefix.
27383 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
27384 (output_indirect_thunk_function): Likewise.
27386 (ix86_code_end): Update output_indirect_thunk_function calls.
27387 (ix86_output_indirect_branch_via_reg): Replace
27388 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
27389 (ix86_output_indirect_branch_via_push): Likewise.
27390 (ix86_output_function_return): Likewise.
27391 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
27392 incompatible with -fcf-protection=branch and
27393 -fcheck-pointer-bounds.
27395 2018-02-22 Steve Ellcey <sellcey@cavium.com>
27398 * config/aarch64/aarch64.c (aarch64_print_address_internal):
27399 Change gcc_assert call to output_operand_lossage.
27401 2018-02-22 Steve Ellcey <sellcey@cavium.com>
27403 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
27405 2018-02-22 DJ Delorie <dj@redhat.com>
27406 Sebastian Perta <sebastian.perta@renesas.com>
27407 Oleg Endo <olegendo@gcc.gnu.org>
27409 * config/rx/rx.c (rx_rtx_costs): New function.
27410 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
27412 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27414 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
27416 2018-02-22 Martin Liska <mliska@suse.cz>
27419 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
27420 Add "native" as a possible value.
27422 2018-02-22 Martin Liska <mliska@suse.cz>
27425 * config/i386/i386.c (ix86_option_override_internal):
27426 Add "native" as a possible value for -march and -mtune.
27428 2018-02-22 Jakub Jelinek <jakub@redhat.com>
27431 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
27432 to all type variants.
27434 PR tree-optimization/84503
27435 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
27436 width as info->bitpos + info->bitsize - start.
27437 (merged_store_group::merge_overlapping): Simplify width computation.
27438 (check_no_overlap): New function.
27439 (imm_store_chain_info::try_coalesce_bswap): Compute expected
27440 start + width and last_order of the group, fail if check_no_overlap
27442 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
27443 to group if check_no_overlap fails.
27445 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
27447 * config/rs6000/altivec.md: Delete contraint arguments to
27448 define_expand, define_split, and define_peephole2, and in
27449 define_insn_and_split if always unused.
27450 * config/rs6000/darwin.md: Ditto.
27451 * config/rs6000/dfp.md: Ditto.
27452 * config/rs6000/rs6000.md: Ditto.
27453 * config/rs6000/sync.md: Ditto.
27454 * config/rs6000/vector.md: Ditto.
27455 * config/rs6000/vsx.md: Ditto.
27457 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
27459 * config/rs6000/altivec.md: Write output control strings as braced
27460 blocks instead of double-quoted strings.
27461 * config/rs6000/darwin.md: Ditto.
27462 * config/rs6000/rs6000.md: Ditto.
27463 * config/rs6000/vector.md: Ditto.
27464 * config/rs6000/vsx.md: Ditto.
27466 2018-02-21 Jason Merrill <jason@redhat.com>
27468 PR c++/84314 - ICE with templates and fastcall attribute.
27469 * attribs.c (build_type_attribute_qual_variant): Remove assert.
27471 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
27473 * ipa-cp.c (determine_versionability): Fix comment typos.
27475 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
27478 * ipa-cp.c (determine_versionability): Do not version functions caling
27481 2018-02-21 Martin Liska <mliska@suse.cz>
27484 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
27485 Add "native" as a possible value.
27486 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
27487 the macro when native cpu detection is available.
27489 2018-02-21 Martin Liska <mliska@suse.cz>
27492 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
27493 Add "native" as a possible value.
27494 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
27495 when native cpu detection is available.
27497 2018-02-21 Jakub Jelinek <jakub@redhat.com>
27498 Martin Sebor <msebor@redhat.com>
27500 PR tree-optimization/84478
27501 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
27503 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
27504 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
27505 support which is conservatively correct, for 2 only stay conservative
27506 for maxlen. Formatting and comment capitalization fixes. Add STRICT
27507 argument to the 2 argument get_range_strlen, adjust 6 arg
27508 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
27510 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
27511 (gimple_fold_builtin_strlen): Pass true as last argument to
27514 2018-02-20 Martin Sebor <msebor@redhat.com>
27516 PR middle-end/84095
27517 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
27518 (builtin_memref::set_base_and_offset): Same. Handle inner references.
27519 (builtin_memref::builtin_memref): Factor out parts into
27520 set_base_and_offset and call it.
27522 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
27524 PR middle-end/84406
27525 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
27526 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
27527 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
27528 search at the associated MODE_INT.
27530 2018-02-20 Jeff Law <law@redhat.com>
27532 PR middle-end/82123
27533 PR tree-optimization/81592
27534 PR middle-end/79257
27535 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
27536 for range data rather than using global data.
27537 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
27538 range data rather than using global data.
27539 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
27540 pass it to children as needed.
27541 (struct directive::fmtresult): Similarly.
27542 (struct directive::set_width): Similarly.
27543 (struct directive::set_precision): Similarly.
27544 (format_integer, format_directive, parse_directive): Similarly.
27545 (format_none): Accept unnamed vr_values parameter.
27546 (format_percent, format_floating, format_character): Similarly.
27547 (format_string, format_plain): Similarly.
27548 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
27549 the EVRP range analyzer for range data rather than using global data.
27550 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
27551 gimple-ssa-evrp-analyze.h
27552 (class sprintf_dom_walker): Add after_dom_children member function.
27553 Add evrp_range_analyzer member.
27554 (sprintf_dom_walker::before_dom_children): Call into the EVRP
27555 range analyzer as needed.
27556 (sprintf_dom_walker::after_dom_children): New member function.
27557 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
27559 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
27560 (evrp_range_analyzer::pop_to_marker): Likewise.
27562 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
27564 PR tree-optimization/84419
27565 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
27566 with the required type if its current type is compatible but
27569 2018-02-20 Jakub Jelinek <jakub@redhat.com>
27571 PR middle-end/82004
27572 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
27573 after vectorization.
27575 2018-02-20 Martin Liska <mliska@suse.cz>
27578 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
27579 possible values if we don't have a hint.
27581 2018-02-20 Martin Liska <mliska@suse.cz>
27585 * final.c (shorten_branches): Build align_tab array with one
27587 * opts.c (finish_options): Add alignment option limit check.
27588 (MAX_CODE_ALIGN): Likewise.
27589 (MAX_CODE_ALIGN_VALUE): Likewise.
27590 * doc/invoke.texi: Document maximum allowed option value for
27591 all -falign-* options.
27593 2018-02-19 Jakub Jelinek <jakub@redhat.com>
27596 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
27597 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
27598 * var-tracking.c (emit_note_insn_var_location): Remove all references
27599 to NOTE_INSN_CALL_ARG_LOCATION.
27600 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
27601 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
27602 Use copy_rtx_if_shared.
27603 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
27604 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
27605 (dwarf2out_var_location): Remove handling of
27606 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
27608 * final.c (final_scan_insn): Remove all references to
27609 NOTE_INSN_CALL_ARG_LOCATION.
27610 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
27611 before dumping final insns.
27612 * except.c (emit_note_eh_region_end): Remove all references to
27613 NOTE_INSN_CALL_ARG_LOCATION.
27614 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
27615 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
27616 * config/arc/arc.c (hwloop_optimize): Likewise.
27617 * config/arm/arm.c (create_fix_barrier): Likewise.
27618 * config/s390/s390.c (s390_chunkify_start): Likewise.
27619 * config/sh/sh.c (find_barrier): Likewise.
27620 * config/i386/i386.c (rest_of_insert_endbranch,
27621 ix86_seh_fixup_eh_fallthru): Likewise.
27622 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
27623 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
27624 * config/frv/frv.c (frv_function_prologue): Likewise.
27625 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
27627 (note_outside_basic_block_p): Remove all references to
27628 NOTE_INSN_CALL_ARG_LOCATION.
27629 * gengtype.c (adjust_field_rtx_def): Likewise.
27630 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
27632 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
27633 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
27636 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
27639 PR tree-optimization/84452
27640 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
27641 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
27644 2018-02-19 Martin Liska <mliska@suse.cz>
27647 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
27649 2018-02-19 Martin Liska <mliska@suse.cz>
27650 Richard Sandiford <richard.sandiford@linaro.org>
27652 PR tree-optimization/82491
27653 * gimple-fold.c (get_base_constructor): Make earlier bail out
27656 2018-02-19 Carl Love <cel@us.ibm.com>
27658 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
27659 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
27661 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
27662 P8V_BUILTIN_VEC_NEG.
27664 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
27666 * config/rl78/rl78.md (movdf): New define expand.
27668 2018-02-19 Martin Liska <mliska@suse.cz>
27671 * doc/invoke.texi: Fix typo.
27672 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
27674 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
27676 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
27677 handle rs6000_single_float and rs6000_double_float specially for
27680 2018-02-16 Jeff Law <law@redhat.com>
27682 * config/rx/rx.c (add_pop_cfi_notes): New function.;
27683 (pop_regs): Use it.
27685 2018-02-16 Jakub Jelinek <jakub@redhat.com>
27688 * ipa-inline.c (inline_small_functions): Fix a typo.
27690 2018-02-16 Nathan Sidwell <nathan@acm.org>
27692 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
27694 2018-02-16 Carl Love <cel@us.ibm.com>
27696 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
27697 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
27698 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
27699 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
27700 expansion to P8V_BUILTIN_VEC_FLOAT2.
27702 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
27704 PR rtl-optimization/70023
27705 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
27706 src_regno into account.
27708 2018-02-16 Carl Love <cel@us.ibm.com>
27710 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
27711 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
27712 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
27713 * config/rs6000/rs6000.c: Remove case statements for
27714 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
27715 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
27716 and P9V_BUILTIN_VEC_VINSERT4B.
27717 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
27718 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
27719 * config/rs6000/vsx.md:
27720 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
27723 2018-02-16 Carl Love <cel@us.ibm.com>
27725 * config/rs6000/altivec.h: Add builtin names vec_extract4b
27727 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
27729 * config/rs6000/rs6000-c.c: Add the definitions for
27730 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
27731 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
27732 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
27733 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
27734 definition for insert4b and define insn *insert3b_internal.
27735 * doc/extend.texi: Add documentation for vec_extract4b.
27737 2018-02-16 Nathan Sidwell <nathan@acm.org>
27739 * doc/extend.texi (Backwards Compatibility): Mention friend
27740 injection. Note for-scope is deprecated.
27741 * doc/invoke.texi (-ffriend-injection): Deprecate.
27743 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
27745 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
27746 that moved to I2, also allow destinations that are a paradoxical
27747 subreg (instead of a normal reg).
27749 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
27752 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
27755 2018-02-16 Richard Biener <rguenther@suse.de>
27757 PR tree-optimization/84037
27758 PR tree-optimization/84016
27760 * config/i386/i386.c (ix86_builtin_vectorization_cost):
27761 Adjust vec_construct for the fact we need additional higher latency
27762 128bit inserts for AVX256 and AVX512 vector builds.
27763 (ix86_add_stmt_cost): Scale vector construction cost for
27766 2018-02-16 Richard Biener <rguenther@suse.de>
27768 PR tree-optimization/84417
27769 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
27770 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
27771 (non_rewritable_lvalue_p): Likewise, use poly-ints.
27773 2018-02-16 Martin Liska <mliska@suse.cz>
27776 * internal-fn.def (ASAN_CHECK): Set proper flags.
27777 (ASAN_MARK): Likewise.
27779 2018-02-16 Julia Koval <julia.koval@intel.com>
27781 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
27782 from PTA_CANNONLAKE.
27784 2018-02-16 Jakub Jelinek <jakub@redhat.com>
27787 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
27788 Use ++iter rather than iter++ for std::list iterators.
27789 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
27790 defer deleting them until all nodes in the forest are processed. Do
27791 free even leaf nodes. Change to_process into auto_vec.
27794 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
27795 * vec.h (vec_default_construct): Use memset instead of placement new
27796 if BROKEN_VALUE_INITIALIZATION is defined.
27797 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
27798 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
27801 PR rtl-optimization/83723
27802 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
27803 * lra.c (lra_substitute_pseudo): Likewise. If true, use
27804 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
27806 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
27808 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
27810 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
27812 PR rtl-optimization/81443
27813 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
27814 from inner REGs to paradoxical SUBREGs.
27816 2018-02-16 Richard Biener <rguenther@suse.de>
27818 PR tree-optimization/84399
27819 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
27820 For operands we can analyze at their definition make sure we can
27821 analyze them at each use as well.
27823 2018-02-16 Richard Biener <rguenther@suse.de>
27825 PR tree-optimization/84190
27826 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
27827 volatile accesses if the decl isn't volatile.
27829 2018-02-15 Jason Merrill <jason@redhat.com>
27831 PR c++/84314 - ICE with templates and fastcall attribute.
27832 * attribs.c (build_type_attribute_qual_variant): Don't clobber
27833 TYPE_CANONICAL on an existing type.
27835 2018-02-15 Jakub Jelinek <jakub@redhat.com>
27837 PR tree-optimization/84383
27838 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
27839 dstoff nor call operand_equal_p if dstbase is NULL.
27841 PR tree-optimization/84334
27842 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
27843 also a CONSTANT_CLASS_P, punt.
27845 2018-02-14 Jim Wilson <jimw@sifive.com>
27847 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
27848 first SMALL_OPERAND check. New local min_second_step. Move assert
27849 to where locals are set. Add TARGET_RVC support.
27850 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
27852 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
27854 * doc/invoke.texi: Correct -Wformat-overflow code sample.
27856 2018-02-14 Martin Sebor <msebor@redhat.com>
27858 PR tree-optimization/83698
27859 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
27860 arrays constrain the offset range to their bounds.
27861 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
27862 (builtin_access::overlap): Avoid setting the size of overlap if it's
27864 (maybe_diag_overlap): Also consider arrays when deciding what values
27865 of offsets to include in diagnostics.
27867 2018-02-14 Martin Sebor <msebor@redhat.com>
27870 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
27871 that correspond to the kind of a declaration.
27873 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
27876 * config/pa/pa.md: Load address of PIC label using the linkage table
27877 if the label is nonlocal.
27879 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
27881 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
27882 warning message if user requests -maltivec=be.
27883 * doc/invoke.texi: Document deprecation of -maltivec=be.
27885 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
27888 * config/rs6000/rs6000-c.c: Update definitions for
27889 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
27890 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
27892 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
27895 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
27896 add _get_ssp intrinsics. Remove argument from
27897 __builtin_ia32_rdssp[d|q].
27898 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
27899 * config/i386/i386-builtin.def: Remove argument from
27900 __builtin_ia32_rdssp[d|q].
27901 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
27902 ix86_expand_special_args_builtin for _rdssp[d|q].
27903 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
27904 Clear register before usage.
27905 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
27906 Add documentation for new _get_ssp and _inc_ssp intrinsics.
27908 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
27910 PR tree-optimization/84357
27911 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
27912 operand 1 of an ARRAY_REF too.
27914 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
27917 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
27918 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
27920 (set_of_reg): New struct.
27921 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
27922 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
27923 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
27925 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
27926 Split into bitclr, bitset, bitinvert patterns if appropriate.
27927 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
27928 use rx_fuse_in_memory_bitop.
27929 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
27930 to named insn, correct maximum insn length.
27932 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
27935 * machmode.def: Define a complex mode for PARTIAL_INT.
27936 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
27938 * doc/rtl.texi: Document CSPImode.
27939 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
27941 (msp430_hard_regno_nregs_with_padding): Likewise.
27943 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
27946 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
27948 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
27950 PR rtl-optimization/84169
27951 * combine.c (try_combine): New variable split_i2i3. Set it to true if
27952 we generated a parallel as new i3 and we split that to new i2 and i3
27953 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
27954 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
27955 those to i2, not i1. Partially rewrite this scan code.
27957 2018-02-13 Jakub Jelinek <jakub@redhat.com>
27960 * stor-layout.c (place_field): For variable length fields, adjust
27961 offset_align afterwards not just based on the field's alignment,
27962 but also on the size.
27964 PR middle-end/84309
27965 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
27966 of exps and logs in the use_exp2 case.
27968 2018-02-13 Jeff Law <law@redhat.com>
27970 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
27971 entry for "vector".
27973 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
27976 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
27980 * common.opt (gas-loc-support, gas-locview-support): New.
27981 (ginline-points, ginternal-reset-location-views): New.
27982 * doc/invoke.texi: Document them. Use @itemx where intended.
27983 (gvariable-location-views): Adjust.
27984 * target.def (reset_location_view): New.
27985 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
27986 (TARGET_RESET_LOCATION_VIEW): New.
27987 * doc/tm.texi: Rebuilt.
27988 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
27989 (dwarf2out_default_as_locview_support): New.
27990 (output_asm_line_debug_info): Use option variables.
27991 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
27992 (output_loc_list): Likewise.
27993 (add_high_low_attributes): Check option variables.
27994 Don't output entry view attribute in strict mode.
27995 (gen_inlined_subroutine_die): Check option variables.
27996 (dwarf2out_inline_entry): Likewise.
27997 (init_sections_and_labels): Likewise.
27998 (dwarf2out_early_finish): Likewise.
27999 (maybe_reset_location_view): New, from...
28000 (dwarf2out_var_location): ... here. Call it.
28001 * debug.h (dwarf2out_default_as_loc_support): Declare.
28002 (dwarf2out_default_as_locview_support): Declare.
28003 * hooks.c (hook_int_rtx_insn_0): New.
28004 * hooks.h (hook_int_rtx_insn_0): Declare.
28005 * toplev.c (process_options): Take -gas-loc-support and
28006 -gas-locview-support from dwarf2out. Enable
28007 -gvariable-location-views by default only with locview
28008 assembler support. Enable -ginternal-reset-location-views by
28009 default only if the target defines the corresponding hook.
28010 Enable -ginline-points by default if location views are
28011 enabled; force it disabled if statement frontiers are
28013 * tree-inline.c (expand_call_inline): Check option variables.
28014 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
28016 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
28018 PR tree-optimization/84321
28019 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
28020 handling. Also check whether the anti-range contains any values
28021 that satisfy the mask; switch to a VR_RANGE if not.
28023 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
28026 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
28028 2018-02-13 Martin Jambor <mjambor@suse.cz>
28031 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
28032 of call statements, also set location of a load to a temporary.
28034 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
28036 * config/rl78/rl78.c (add_vector_labels): New function.
28037 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
28038 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
28039 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
28040 which checks that no arguments are passed.
28041 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
28042 * doc/extend.texi: Documentation for the new attribute.
28044 2018-02-13 Andreas Schwab <schwab@suse.de>
28046 * config/riscv/linux.h (CPP_SPEC): Define.
28048 2018-02-13 Jakub Jelinek <jakub@redhat.com>
28051 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
28052 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
28053 OPTION_MASK_ISA_AES as first argument to def_builtin_const
28054 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
28055 instead of OPTION_MASK_ISA_PCLMUL as first argument to
28056 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
28057 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
28058 temporarily for AES and PCLMUL builtins.
28060 PR tree-optimization/84339
28061 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
28062 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
28065 PR middle-end/84309
28066 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
28067 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
28068 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
28070 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
28072 * omp-simd-clone.h: New file.
28073 * omp-simd-clone.c: Include omp-simd-clone.h.
28074 (expand_simd_clones): No longer static.
28075 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
28076 cgraph.h and omp-simd-clone.h.
28077 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
28078 (vect_recog_widen_shift_pattern): Formatting fix.
28079 (vect_pattern_recog_1): Don't check optab for calls.
28082 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
28083 operands[2] into a REG before using gen_lowpart on it.
28085 2018-02-12 Jeff Law <law@redhat.com>
28088 * config/sh/sh.c (find_barrier): Consider a sibling call
28091 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
28092 successfully back substituting a reg.
28094 2018-02-12 Richard Biener <rguenther@suse.de>
28096 PR tree-optimization/84037
28097 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
28098 parameter, move visited init to caller.
28099 (vect_slp_analyze_operations): Separate cost from validity
28100 check, initialize visited once for all instances.
28101 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
28103 * tree-vect-stmts.c (vect_model_simple_cost): Make early
28105 (vect_model_promotion_demotion_cost): Likewise.
28106 (vectorizable_bswap): Guard cost modeling with !slp_node
28107 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
28109 (vectorizable_call): Likewise.
28110 (vectorizable_conversion): Likewise.
28111 (vectorizable_assignment): Likewise.
28112 (vectorizable_shift): Likewise.
28113 (vectorizable_operation): Likewise.
28114 (vectorizable_store): Likewise.
28115 (vectorizable_load): Likewise.
28116 (vectorizable_condition): Likewise.
28117 (vectorizable_comparison): Likewise.
28119 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
28122 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
28123 (ASAN_MARK): Fix fnspec to account for return value, change pointer
28124 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
28126 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
28128 PR middle-end/83665
28129 * params.def (inline-min-speedup): Increase from 8 to 15.
28130 (max-inline-insns-auto): Decrease from 40 to 30.
28131 * ipa-split.c (consider_split): Add some buffer for function to
28132 be considered inlining candidate.
28133 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
28136 2018-02-12 Richard Biener <rguenther@suse.de>
28138 PR tree-optimization/84037
28139 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
28140 matched stmts if we cannot swap the non-matched ones.
28142 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
28144 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
28145 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
28146 _mm_maskz_scalef_round_ss): New intrinsics.
28147 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
28148 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
28149 __builtin_ia32_scalefss_round): Remove.
28150 (__builtin_ia32_scalefsd_mask_round,
28151 __builtin_ia32_scalefss_mask_round): New intrinsics.
28152 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
28153 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
28154 ((match_operand:VF_128 2 "<round_nimm_predicate>"
28155 "<round_constraint>")): Changed to ...
28156 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
28157 "<round_scalar_constraint>")): ... this.
28158 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
28159 %0, %1, %2<round_op3>}"): Changed to ...
28160 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
28161 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
28162 %2<round_scalar_mask_op3>}"): ... this.
28163 * config/i386/subst.md (round_scalar_nimm_predicate): New.
28165 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
28167 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
28168 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
28169 (_mm_maskz_sqrt_round_ss): New intrinsics.
28170 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
28171 (__builtin_ia32_sqrtsd_mask_round)
28172 (__builtin_ia32_sqrtss_mask_round): New builtins.
28173 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
28174 (__builtin_ia32_sqrtss_round): Remove.
28175 (__builtin_ia32_sqrtsd_mask_round)
28176 (__builtin_ia32_sqrtss_mask_round): New builtins.
28177 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
28178 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
28179 ((match_operand:VF_128 1 "vector_operand"
28180 "xBm,<round_constraint>")): Changed to ...
28181 ((match_operand:VF_128 1 "vector_operand"
28182 "xBm,<round_scalar_constraint>")): ... this.
28183 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
28184 %0, %2, %<iptr>1<round_op3>}): Changed to ...
28185 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
28186 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
28187 %<iptr>1<round_scalar_mask_op3>}): ... this.
28188 ((set_attr "prefix" "<round_prefix>")): Changed to ...
28189 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
28191 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
28194 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
28195 Cast vec_cmpeq result to correct type.
28196 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
28197 Cast vec_cmpgt result to correct type.
28199 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
28201 * final.c (final_scan_insn_1): Renamed from...
28202 (final_scan_insn): ... this. New wrapper, to recover
28203 seen from the outermost call in recursive ones.
28204 * config/sparc/sparc.c (output_return): Drop seen from call.
28205 (output_sibcall): Likewise.
28206 * config/visium/visium.c (output_branch): Likewise.
28208 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
28210 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
28213 2018-02-10 Alan Modra <amodra@gmail.com>
28216 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
28217 Specify LR as an input.
28219 2018-02-10 Jakub Jelinek <jakub@redhat.com>
28222 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
28223 remove_member_access_dummy_vars): New functions.
28224 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
28225 lower_omp_1, execute_lower_omp): Use them.
28227 PR rtl-optimization/84308
28228 * shrink-wrap.c (spread_components): Release todo vector.
28230 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
28232 PR rtl-optimization/57193
28233 * ira-color.c (struct allocno_color_data): Add member
28234 conflict_allocno_hard_prefs.
28235 (update_conflict_allocno_hard_prefs): New.
28236 (bucket_allocno_compare_func): Add a preference based on
28237 conflict_allocno_hard_prefs.
28238 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
28239 (color_allocnos): Remove a dead code. Initiate
28240 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
28242 2018-02-09 Jakub Jelinek <jakub@redhat.com>
28245 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
28246 constraint from =wa to wa. Avoid a subreg on the output operand,
28247 instead use a pseudo and subreg it in a move.
28248 (p9_xxbrd_<mode>): Changed to ...
28249 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
28250 (p9_xxbrd_v2df): New expander.
28251 (p9_xxbrw_<mode>): Changed to ...
28252 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
28253 (p9_xxbrw_v4sf): New expander.
28255 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
28257 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
28259 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
28262 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
28263 multiply in 32-bit mode.
28264 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
28265 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
28268 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
28270 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
28271 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
28272 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
28273 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
28275 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
28278 * dwarf2out.c (is_trivial_indirect_ref): New function.
28279 (dwarf2out_late_global_decl): Do not generate a location
28280 attribute for variables that have a non-trivial DECL_VALUE_EXPR
28281 and that are not defined in the current unit.
28283 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
28285 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
28286 instead of a libcall for UNORDERED.
28288 2018-02-09 Tamar Christina <tamar.christina@arm.com>
28291 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
28292 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
28294 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28297 * config/s390/s390.c (s390_set_current_function): Invoke
28298 s390_indirect_branch_settings also if fndecl didn't change.
28300 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
28302 * config/rs6000/rs6000.md (blockage): Set length to zero.
28304 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
28306 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
28308 2018-02-09 Jakub Jelinek <jakub@redhat.com>
28311 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
28312 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
28316 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
28317 PARALLEL incoming that failed vt_get_decl_and_offset check.
28319 PR middle-end/84237
28320 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
28321 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
28323 (get_variable_section): For decls in named .bss* sections pass true as
28324 second argument to bss_initializer_p.
28326 2018-02-09 Marek Polacek <polacek@redhat.com>
28327 Jakub Jelinek <jakub@redhat.com>
28330 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
28331 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
28332 Sync some changes from cxx_fold_indirect_ref.
28334 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
28336 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
28338 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
28339 (BLOCK_INLINE_ENTRY_LABEL): New.
28340 (dwarf2out_var_location): Disregard inline entry markers.
28341 (inline_entry_data): New struct.
28342 (inline_entry_data_hasher): New hashtable type.
28343 (inline_entry_data_hasher::hash): New.
28344 (inline_entry_data_hasher::equal): New.
28345 (inline_entry_data_table): New variable.
28346 (add_high_low_attributes): Add DW_AT_entry_pc and
28347 DW_AT_GNU_entry_view attributes if a pending entry is found
28348 in inline_entry_data_table. Add old entry_pc attribute only
28349 if debug nonbinding markers are disabled.
28350 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
28351 markers are enabled.
28352 (block_within_block_p, dwarf2out_inline_entry): New.
28353 (dwarf2out_finish): Check that no entries remained in
28354 inline_entry_data_table.
28355 * final.c (reemit_insn_block_notes): Handle inline entry notes.
28356 (final_scan_insn, notice_source_line): Likewise.
28357 (rest_of_clean_state): Skip inline entry markers.
28358 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
28360 * gimple.c (gimple_build_debug_inline_entry): New.
28361 * gimple.h (enum gimple_debug_subcode): Add
28362 GIMPLE_DEBUG_INLINE_ENTRY.
28363 (gimple_build_debug_inline_entry): Declare.
28364 (gimple_debug_inline_entry_p): New.
28365 (gimple_debug_nonbind_marker_p): Adjust.
28366 * insn-notes.def (INLINE_ENTRY): New.
28367 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
28368 inline entry marker notes.
28369 (print_insn): Likewise.
28370 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
28371 (INSN_DEBUG_MARKER_KIND): Likewise.
28372 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
28373 * tree-inline.c (expand_call_inline): Build and insert
28374 debug_inline_entry stmt.
28375 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
28376 inline entry blocks early, if nonbind markers are enabled.
28377 (dump_scope_block): Dump fragment info.
28378 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
28379 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
28380 (gimple_build_debug_inline_entry): New.
28381 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
28382 Enable/disable inline entry points too.
28383 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
28384 (DEBUG_INSN): Describe inline entry markers.
28386 * common.opt (gvariable-location-views): New.
28387 (gvariable-location-views=incompat5): New.
28388 * config.in: Rebuilt.
28389 * configure: Rebuilt.
28390 * configure.ac: Test assembler for view support.
28391 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
28392 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
28393 * dwarf2out.c (var_loc_view): New typedef.
28394 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
28395 (dwarf2out_locviews_in_attribute): New.
28396 (dwarf2out_locviews_in_loclist): New.
28397 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
28398 (enum dw_line_info_opcode): Add LI_adv_address.
28399 (struct dw_line_info_table): Add view.
28400 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
28401 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
28402 (zero_view_p): New variable.
28403 (ZERO_VIEW_P): New macro.
28404 (output_asm_line_debug_info): New.
28405 (struct var_loc_node): Add view.
28406 (add_AT_view_list, AT_loc_list): New.
28407 (add_var_loc_to_decl): Add view param. Test it against last.
28408 (new_loc_list): Add view params. Record them.
28409 (AT_loc_list_ptr): Handle loc and view lists.
28410 (view_list_to_loc_list_val_node): New.
28411 (print_dw_val): Handle dw_val_class_view_list.
28412 (size_of_die): Likewise.
28413 (value_format): Likewise.
28414 (loc_list_has_views): New.
28415 (gen_llsym): Set vl_symbol too.
28416 (maybe_gen_llsym, skip_loc_list_entry): New.
28417 (dwarf2out_maybe_output_loclist_view_pair): New.
28418 (output_loc_list): Output view list or entries too.
28419 (output_view_list_offset): New.
28420 (output_die): Handle dw_val_class_view_list.
28421 (output_dwarf_version): New.
28422 (output_compilation_unit_header): Use it.
28423 (output_skeleton_debug_sections): Likewise.
28424 (output_rnglists, output_line_info): Likewise.
28425 (output_pubnames, output_aranges): Update version comments.
28426 (output_one_line_info_table): Output view numbers in asm comments.
28427 (dw_loc_list): Determine current endview, pass it to new_loc_list.
28428 Call maybe_gen_llsym.
28429 (loc_list_from_tree_1): Adjust.
28430 (add_AT_location_description): Create view list attribute if
28431 needed, check it's absent otherwise.
28432 (convert_cfa_to_fb_loc_list): Adjust.
28433 (maybe_emit_file): Call output_asm_line_debug_info for test.
28434 (dwarf2out_var_location): Reset views as needed. Precompute
28435 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
28436 attribute. Set view.
28437 (new_line_info_table): Reset next view.
28438 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
28439 (dwarf2out_source_line): Likewise. Output view resets and labels to
28440 the assembler, or select appropriate line info opcodes.
28441 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
28442 (optimize_string_length): Catch it. Adjust.
28443 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
28444 dw_val_class_view_list, and remove it if no longer needed.
28445 (hash_loc_list): Hash view numbers.
28446 (loc_list_hasher::equal): Compare them.
28447 (optimize_location_lists): Check whether a view list symbol is
28448 needed, and whether the locview attribute is present, and
28449 whether they match. Remove the locview attribute if no longer
28451 (index_location_lists): Call skip_loc_list_entry for test.
28452 (dwarf2out_finish): Call output_asm_line_debug_info for test.
28453 Use output_dwarf_version.
28454 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
28455 (struct dw_val_node): Add val_view_list.
28456 * final.c (SEEN_NEXT_VIEW): New.
28457 (set_next_view_needed): New.
28458 (clear_next_view_needed): New.
28459 (maybe_output_next_view): New.
28460 (final_start_function): Rename to...
28461 (final_start_function_1): ... this. Take pointer to FIRST,
28462 add SEEN parameter. Emit param bindings in the initial view.
28463 (final_start_function): Reintroduce SEEN-less interface.
28464 (final): Rename to...
28465 (final_1): ... this. Take SEEN parameter. Output final pending
28466 next view at the end.
28467 (final): Reintroduce seen-less interface.
28468 (final_scan_insn): Output pending next view before switching
28469 sections or ending a block. Mark the next view as needed when
28470 outputting variable locations. Notify debug backend of section
28471 changes, and of location view changes.
28472 (rest_of_handle_final): Adjust.
28473 * toplev.c (process_options): Autodetect value for debug variable
28474 location views option. Warn on incompat5 without -gdwarf-5.
28475 * doc/invoke.texi (gvariable-location-views): New.
28476 (gvariable-location-views=incompat5): New.
28477 (gno-variable-location-views): New.
28479 2018-02-08 David Malcolm <dmalcolm@redhat.com>
28481 PR tree-optimization/84136
28482 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
28483 that the result of find_edge is non-NULL.
28485 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
28488 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
28489 storing integer register in SImode. Fix cost of 256 and 512
28490 byte aligned SSE register store.
28492 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
28494 * config/i386/i386.c (ix86_multiplication_cost): Fix
28495 multiplication cost for TARGET_AVX512DQ.
28497 2018-02-08 Marek Polacek <polacek@redhat.com>
28499 PR tree-optimization/84238
28500 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
28503 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
28505 PR tree-optimization/84265
28506 * tree-vect-stmts.c (vectorizable_store): Don't treat
28507 VMAT_CONTIGUOUS accesses as grouped.
28508 (vectorizable_load): Likewise.
28510 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
28512 PR tree-optimization/81635
28513 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
28514 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
28515 (test_round_for_mask): New functions.
28516 (wide_int_cc_tests): Call test_round_for_mask.
28517 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
28518 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
28519 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
28520 range returned by get_range_info.
28522 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
28525 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
28526 * symtab.c: Include builtins.h
28527 (symtab_node::output_to_lto_symbol_table_p): Move here
28528 from lto-streamer-out.c:output_symbol_p.
28529 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
28530 (output_symbol_p): Move all logic to symtab.c
28531 (produce_symtab): Update.
28533 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28535 * config/s390/s390-opts.h (enum indirect_branch): Define.
28536 * config/s390/s390-protos.h (s390_return_addr_from_memory)
28537 (s390_indirect_branch_via_thunk)
28538 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
28539 (enum s390_indirect_branch_type): Define.
28540 * config/s390/s390.c (struct s390_frame_layout, struct
28541 machine_function): Remove.
28542 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
28543 (indirect_branch_table_label_no, indirect_branch_table_name):
28545 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
28546 (enum s390_indirect_branch_option): Define.
28547 (s390_return_addr_from_memory): New function.
28548 (s390_handle_string_attribute): New function.
28549 (s390_attribute_table): Add new attribute handler.
28550 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
28551 (s390_indirect_branch_via_thunk): New function.
28552 (s390_indirect_branch_via_inline_thunk): New function.
28553 (s390_function_ok_for_sibcall): When jumping via thunk disallow
28554 sibling call optimization for non z10 compiles.
28555 (s390_emit_call): Force indirect branch target to be a single
28556 register. Add r1 clobber for non-z10 compiles.
28557 (s390_emit_epilogue): Emit return jump via return_use expander.
28558 (s390_reorg): Handle JUMP_INSNs as execute targets.
28559 (s390_option_override_internal): Perform validity checks for the
28560 new command line options.
28561 (s390_indirect_branch_attrvalue): New function.
28562 (s390_indirect_branch_settings): New function.
28563 (s390_set_current_function): Invoke s390_indirect_branch_settings.
28564 (s390_output_indirect_thunk_function): New function.
28565 (s390_code_end): Implement target hook.
28566 (s390_case_values_threshold): Implement target hook.
28567 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
28569 * config/s390/s390.h (struct s390_frame_layout)
28570 (struct machine_function): Move here from s390.c.
28571 (TARGET_INDIRECT_BRANCH_NOBP_RET)
28572 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
28573 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
28574 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
28575 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
28576 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
28577 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
28578 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
28579 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
28580 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
28581 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
28582 (mnemonic attribute): Add values which aren't recognized
28584 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
28585 pattern for branch conversion. Fix mnemonic attribute.
28586 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
28587 indirect branch via thunk if requested.
28588 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
28589 ("*indirect_jump"): Disable for branch conversion using out of
28591 ("indirect_jump_via_thunk<mode>_z10")
28592 ("indirect_jump_via_thunk<mode>")
28593 ("indirect_jump_via_inlinethunk<mode>_z10")
28594 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
28595 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
28596 ("casesi_jump_via_inlinethunk<mode>_z10")
28597 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
28598 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
28599 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
28600 ("*indirect2_jump"): Disable for branch conversion.
28601 ("casesi_jump"): Turn into expander and expand patterns for branch
28603 ("return_use"): New expander.
28604 ("*return"): Emit return via thunk and rename it to ...
28605 ("*return<mode>"): ... this one.
28606 * config/s390/s390.opt: Add new options and and enum for the
28609 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
28611 * lra-constraints.c (match_reload): Unconditionally use
28612 gen_lowpart_SUBREG, rather than selecting between that
28613 and equivalent gen_rtx_SUBREG code.
28615 2018-02-08 Richard Biener <rguenther@suse.de>
28617 PR tree-optimization/84233
28618 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
28619 changed flag instead of boguously re-using phi_inserted.
28621 2018-02-08 Martin Jambor <mjambor@suse.cz>
28623 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
28624 static local variables.
28626 2018-02-08 Richard Biener <rguenther@suse.de>
28628 PR tree-optimization/84278
28629 * tree-vect-stmts.c (vectorizable_store): When looking for
28630 smaller vector types to perform grouped strided loads/stores
28631 make sure the mode is supported by the target.
28632 (vectorizable_load): Likewise.
28634 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
28636 * config/aarch64/aarch64.c (aarch64_components_for_bb):
28637 Increase LDP/STP opportunities by adding adjacent callee-saves.
28639 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
28641 PR rtl-optimization/84068
28642 PR rtl-optimization/83459
28643 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
28645 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
28647 PR tree-optimization/84224
28648 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
28649 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
28650 non-zero arguments.
28652 2018-02-07 Iain Sandoe <iain@codesourcery.com>
28655 * config/rs6000/altivec.md (*restore_world): Remove LR use.
28656 * config/rs6000/predicates.md (restore_world_operation): Adjust op
28657 count, remove one USE.
28659 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
28661 * doc/install.texi (Configuration): Document the
28662 --with-long-double-format={ibm,ieee} PowerPC configuration
28666 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
28667 Convert from define_expand to be define_insn_and_split. Rework
28668 float/double/_Float128 conversions to QI/HI/SImode to work with
28669 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
28670 conversions to QI/HImode types did a store and then a load to
28671 truncate the value. For conversions to VSX registers, don't split
28672 the insn, instead emit the code directly. Use the code iterator
28673 any_fix to combine signed and unsigned conversions.
28674 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
28675 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
28676 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
28677 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
28678 (fix_<mode>di2_hw): Likewise.
28679 (fixuns_<mode>di2_hw): Likewise.
28680 (fix_<mode>si2_hw): Likewise.
28681 (fixuns_<mode>si2_hw): Likewise.
28682 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
28683 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
28684 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
28685 fix<uns>_trunc<SFDF:mode>si2_p8.
28686 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
28687 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
28688 (fix<uns>_<mode>_mem): Likewise.
28689 (fctiw<u>z_<mode>_mem): Likewise.
28690 (fix<uns>_<mode>_mem): Likewise.
28691 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
28692 the register allocator from doing a direct move to the GPRs to do
28693 a store, and instead use the ISA 3.0 store byte/half-word from
28694 vector register instruction. For IEEE 128-bit floating point,
28695 also optimize stores of 32-bit ints.
28696 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
28698 2018-02-07 Alan Hayward <alan.hayward@arm.com>
28700 * genextract.c (push_pathstr_operand): New function to support
28702 (walk_rtx): Call push_pathstr_operand.
28703 (print_path): Support [a-zA-Z].
28705 2018-02-07 Richard Biener <rguenther@suse.de>
28707 PR tree-optimization/84037
28708 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
28709 (cse_and_gimplify_to_preheader): Declare.
28710 (vect_get_place_in_interleaving_chain): Likewise.
28711 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
28713 (_loop_vec_info::~_loop_vec_info): Delete it.
28714 (cse_and_gimplify_to_preheader): New function.
28715 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
28716 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
28717 (vectorizable_load): Likewise. For grouped stores always base
28718 the IV on the first element.
28719 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
28720 condition before gimplifying.
28722 2018-02-07 Jakub Jelinek <jakub@redhat.com>
28724 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
28725 *DIV_EXPR and *MOD_EXPR.
28727 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
28730 * config/i386/i386.c (ix86_option_override_internal): Mask out
28731 the CF_SET bit when checking -fcf-protection.
28733 2018-02-07 Tom de Vries <tom@codesourcery.com>
28736 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
28739 2018-02-07 Richard Biener <rguenther@suse.de>
28741 PR tree-optimization/84204
28742 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
28745 PR tree-optimization/84205
28746 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
28747 special-case isl_ast_op_zdiv_r.
28749 PR tree-optimization/84223
28750 * graphite-scop-detection.c (gather_bbs::before_dom_children):
28751 Only add conditions from within the region.
28752 (gather_bbs::after_dom_children): Adjust.
28754 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
28757 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
28758 * config/avr/avr.md: Only post-reload split REG-REG moves if
28759 either register is GENERAL_REG_P.
28761 2018-02-07 Jakub Jelinek <jakub@redhat.com>
28763 PR tree-optimization/84235
28764 * tree-ssa-scopedtables.c
28765 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
28766 if the subtraction is performed in floating point type where NaNs are
28767 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
28768 build 1. Formatting fix.
28770 2018-02-06 Jakub Jelinek <jakub@redhat.com>
28773 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
28774 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
28775 and skip it regardless of bb boundaries. Use CALL_P macro,
28776 don't test INSN_P (insn) together with CALL_P or JUMP_P check
28777 unnecessarily, formatting fix.
28779 2018-02-06 Michael Collison <michael.collison@arm.com>
28781 * config/arm/thumb2.md:
28782 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
28783 (*thumb_mov_notscc): Ditto.
28785 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
28788 * config/rs6000/rs6000.md (su code attribute): Use "u" for
28789 unsigned_fix, not "s".
28791 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28793 * configure.ac (gcc_fn_eh_frame_ro): New function.
28794 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
28795 correct .eh_frame permissions.
28796 * configure: Regenerate.
28798 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
28800 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
28801 irrelevant options.
28803 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28805 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28806 Display warning message for -mno-speculate-indirect-jumps.
28808 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
28810 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
28812 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
28814 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
28816 PR tree-optimization/84225
28817 * tree-eh.c (find_trapping_overflow): Only call
28818 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
28820 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
28823 * config/i386/i386.c: Reimplement the check of possible options
28824 -mibt/-mshstk conbination. Change error messages.
28825 * doc/invoke.texi: Fix a typo: remove extra '='.
28827 2018-02-06 Marek Polacek <polacek@redhat.com>
28829 PR tree-optimization/84228
28830 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
28832 2018-02-06 Tamar Christina <tamar.christina@arm.com>
28835 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
28836 emitted arch directives.
28837 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
28838 __ARM_FEATURE_COPROC before changing architectures.
28840 2018-02-06 Richard Biener <rguenther@suse.de>
28842 * config/i386/i386.c (print_reg): Fix typo.
28843 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
28845 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
28847 * configure: Regenerate.
28849 2018-02-05 Martin Sebor <msebor@redhat.com>
28851 PR tree-optimization/83369
28852 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
28855 2018-02-05 Martin Liska <mliska@suse.cz>
28857 * doc/invoke.texi: Cherry-pick upstream r323995.
28859 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
28861 * ira.c (ira_init_register_move_cost): Adjust comment.
28863 2018-02-05 Martin Liska <mliska@suse.cz>
28865 PR gcov-profile/84137
28866 * doc/gcov.texi: Fix typo in documentation.
28868 2018-02-05 Martin Liska <mliska@suse.cz>
28870 PR gcov-profile/83879
28871 * doc/gcov.texi: Document necessity of --dynamic-list-data when
28872 using dlopen functionality.
28874 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
28876 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
28877 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
28878 _mm_maskz_range_ss, _mm_mask_range_round_ss,
28879 _mm_maskz_range_round_ss): New intrinsics.
28880 (__builtin_ia32_rangesd128_round)
28881 (__builtin_ia32_rangess128_round): Remove.
28882 (__builtin_ia32_rangesd128_mask_round,
28883 __builtin_ia32_rangess128_mask_round): New builtins.
28884 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
28885 __builtin_ia32_rangess128_round): Remove.
28886 (__builtin_ia32_rangesd128_mask_round,
28887 __builtin_ia32_rangess128_mask_round): New builtins.
28888 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
28889 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
28890 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
28891 "<round_saeonly_constraint>")): Changed to ...
28892 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
28893 "<round_saeonly_scalar_constraint>")): ... this.
28894 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
28895 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
28896 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
28897 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
28898 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
28900 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
28902 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
28904 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
28905 Remove all values except native, 8540 and 8548.
28907 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
28909 * config/i386/i386.c (ix86_output_function_return): Pass
28910 INVALID_REGNUM, instead of -1, as invalid register number to
28911 indirect_thunk_name and output_indirect_thunk.
28913 2018-02-02 Julia Koval <julia.koval@intel.com>
28915 * config.gcc: Add -march=icelake.
28916 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
28917 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
28918 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
28919 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
28920 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
28921 (processor_target_table): Add icelake.
28922 (ix86_option_override_internal): Handle new PTAs.
28923 (get_builtin_code_for_version): Handle icelake.
28924 (M_INTEL_COREI7_ICELAKE): New.
28925 (fold_builtin_cpu): Handle icelake.
28926 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
28927 * doc/invoke.texi: Add -march=icelake.
28929 2018-02-02 Julia Koval <julia.koval@intel.com>
28931 * config/i386/i386.c (ix86_option_override_internal): Change flags type
28932 to wide_int_bitmask.
28933 * wide-int-bitmask.h: New.
28935 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
28938 * config/i386/i386.md: Replace Pmode with word_mode in
28939 builtin_setjmp_setup and builtin_longjmp to support x32.
28941 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
28945 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
28947 (rs6000_supported_cpu_names): New static variable.
28948 (linux_cpu_translation_table): Likewise.
28949 (elf_platform) <cpu>: Define new static variable and use it.
28950 Translate kernel AT_PLATFORM name to canonical name if needed.
28951 Error if platform name is unknown.
28953 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
28956 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
28958 2018-02-01 Jeff Law <law@redhat.com>
28961 * config/i386/i386.c (release_scratch_register_on_entry): Add new
28962 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
28963 the scratch if RELEASE_VIA_POP is false.
28964 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
28965 If we have to save a temporary register, decrement SIZE appropriately.
28966 Pass new arguments to release_scratch_register_on_entry.
28967 (ix86_adjust_stack_and_probe): Likewise.
28968 (ix86_emit_probe_stack_range): Pass new arguments to
28969 release_scratch_register_on_entry.
28971 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
28973 PR rtl-optimization/84157
28974 * combine.c (change_zero_ext): Use REG_P predicate in
28975 front of HARD_REGISTER_P predicate.
28977 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
28979 * config/avr/avr.c (avr_option_override): Move disabling of
28980 -fdelete-null-pointer-checks to...
28981 * common/config/avr/avr-common.c (avr_option_optimization_table):
28984 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28986 PR tree-optimization/81635
28987 * tree-data-ref.c (split_constant_offset_1): For types that
28988 wrap on overflow, try to use range info to prove that wrapping
28991 2018-02-01 Renlin Li <renlin.li@arm.com>
28994 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
28995 TAILCALL_ADDR_REGS.
28996 (aarch64_register_move_cost): Likewise.
28997 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
28998 TAILCALL_ADDR_REGS.
28999 (REG_CLASS_NAMES): Likewise.
29000 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
29001 TAILCALL_ADDR_REGS. Remove IP registers.
29002 * config/aarch64/aarch64.md (Ucs): Update register constraint.
29004 2018-02-01 Richard Biener <rguenther@suse.de>
29006 * domwalk.h (dom_walker::dom_walker): Add additional constructor
29007 for specifying RPO order and allow NULL for that.
29008 * domwalk.c (dom_walker::dom_walker): Likewise.
29009 (dom_walker::walk): Handle NULL RPO order.
29010 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
29012 (rewrite_update_dom_walker): Likewise.
29013 (mark_def_dom_walker): Likewise.
29015 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
29017 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
29018 (aarch64_maybe_expand_sve_subreg_move): Declare.
29019 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
29020 * config/aarch64/predicates.md (aarch64_any_register_operand): New
29022 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
29023 that are semantically a reverse operation.
29024 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
29025 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
29026 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
29028 (aarch64_can_change_mode_class): For big-endian, forbid changes
29029 between two SVE modes if they have different element sizes.
29031 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
29033 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
29034 the TImode handling for big-endian targets.
29036 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
29038 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
29039 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
29041 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
29042 Remove BSWAP handing for big-endian targets and use the form of
29043 LD1RQ appropariate for the mode.
29045 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
29047 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
29048 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
29049 duplicated element.
29051 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
29054 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
29055 check for operands that need to go through aarch64_sve_reload_be.
29057 2018-02-01 Jakub Jelinek <jakub@redhat.com>
29059 PR tree-optimization/81661
29060 PR tree-optimization/84117
29061 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
29062 * tree-eh.c: Include gimplify.h.
29063 (find_trapping_overflow, replace_trapping_overflow,
29064 rewrite_to_non_trapping_overflow): New functions.
29065 * tree-vect-loop.c: Include tree-eh.h.
29066 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
29067 * tree-data-ref.c: Include tree-eh.h.
29068 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
29070 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
29072 PR rtl-optimization/84123
29073 * combine.c (change_zero_ext): Check if hard register satisfies
29074 can_change_dest_mode before calling gen_lowpart_SUBREG.
29076 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
29079 * ira.c (ira_init_register_move_cost): Remove assert.
29081 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
29083 PR rtl-optimization/84071
29084 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
29085 * doc/tm.texi: Regenerate.
29087 2018-01-31 Richard Biener <rguenther@suse.de>
29089 PR tree-optimization/84132
29090 * tree-data-ref.c (analyze_miv_subscript): Properly
29091 check whether evolution_function_is_affine_multivariate_p
29092 before calling gcd_of_steps_may_divide_p.
29094 2018-01-31 Julia Koval <julia.koval@intel.com>
29097 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29098 * config/i386/i386.md (rdpid_rex64) New.
29099 (rdpid): Make 32bit only.
29101 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
29104 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
29105 an IDENTIFIER_NODE for FUNCTION_TYPE's.
29107 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
29110 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
29112 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
29114 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
29116 PR rtl-optimization/84071
29117 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
29118 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
29120 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
29122 * config/arc/arc.c (arc_handle_aux_attribute): New function.
29123 (arc_attribute_table): Add 'aux' attribute.
29124 (arc_in_small_data_p): Consider aux like variables.
29125 (arc_is_aux_reg_p): New function.
29126 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
29127 (arc_get_aux_arg): New function.
29128 (prepare_move_operands): Handle aux-register access.
29129 (arc_handle_aux_attribute): New function.
29130 * doc/extend.texi (ARC Variable attributes): Add subsection.
29132 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
29134 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
29135 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
29136 (arc_attribute_table): Add 'uncached' attribute.
29137 (arc_print_operand): Print '.di' flag for uncached memory
29139 (arc_in_small_data_p): Do not consider for small data the uncached
29141 (arc_is_uncached_mem_p): New function.
29142 * config/arc/predicates.md (compact_store_memory_operand): Check
29143 for uncached memory accesses.
29144 (nonvol_nonimm_operand): Likewise.
29145 * doc/extend.texi (ARC Type Attribute): New subsection.
29147 2018-01-31 Jakub Jelinek <jakub@redhat.com>
29150 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
29151 falign-loops=): Add Optimization flag.
29153 2018-01-30 Jeff Law <law@redhat.com>
29156 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
29157 INT_REGISTERS_SAVED. Check it prior to calling
29158 get_scratch_register_on_entry.
29159 (ix86_adjust_stack_and_probe): Similarly.
29160 (ix86_emit_probe_stack_range): Similarly.
29161 (ix86_expand_prologue): Corresponding changes.
29163 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29166 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
29167 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
29169 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
29172 * lra-constraints.c (curr_insn_transform): Process AND in the
29175 2018-01-30 Jakub Jelinek <jakub@redhat.com>
29177 PR rtl-optimization/83986
29178 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
29179 dependence against last_pending_memory_flush in addition to
29180 pending_jump_insns.
29182 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
29184 PR tree-optimization/81611
29185 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
29188 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29191 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
29194 2018-01-30 Richard Biener <rguenther@suse.de>
29195 Jakub Jelinek <jakub@redhat.com>
29197 PR tree-optimization/84111
29198 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
29199 inner loops added during recursion, as they don't have up-to-date
29202 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
29205 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
29206 (can_inline_edge_by_limits_p): ... here.
29207 (can_early_inline_edge_p, check_callers,
29208 update_caller_keys, update_callee_keys, recursive_inlining,
29209 add_new_edges_to_heap, speculation_useful_p,
29210 inline_small_functions,
29211 inline_small_functions, flatten_function,
29212 inline_to_all_callers_1): Update.
29214 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
29216 * profile-count.c (profile_count::combine_with_ipa_count): Handle
29219 2018-01-30 Richard Biener <rguenther@suse.de>
29221 PR tree-optimization/83008
29222 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
29223 invariant and constant vector uses in stmts when they need
29224 more than one stmt.
29226 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29229 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
29230 * configure: Regenerate.
29232 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
29234 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
29236 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
29237 Use gen_rtx_REG rather than gen_lowpart.
29239 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
29241 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
29242 rather than 0 when creating partial subregs.
29244 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
29246 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
29249 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
29252 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
29253 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
29254 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
29255 flags. This restores the settings used before the 2017-07-24.
29256 Turning off pre increment/decrement/modify allows IVOPTS to
29257 optimize DF/SF loops where the index is an int.
29259 2018-01-29 Richard Biener <rguenther@suse.de>
29260 Kelvin Nilsen <kelvin@gcc.gnu.org>
29263 * tree-vect-stmts.c (vectorizable_call): Don't call
29264 targetm.vectorize_builtin_md_vectorized_function if callee is
29267 2018-01-22 Carl Love <cel@us.ibm.com>
29269 * doc/extend.tex: Fix typo in second arg in
29270 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
29272 2018-01-29 Richard Biener <rguenther@suse.de>
29274 PR tree-optimization/84086
29275 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
29276 (flush_ssaname_freelist): When SSA names were released reset
29277 the SCEV hash table.
29279 2018-01-29 Richard Biener <rguenther@suse.de>
29281 PR tree-optimization/84057
29282 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
29283 removed paths when removing edges.
29285 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
29287 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
29288 -mfunction-return=@var{choice}.
29290 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
29292 PR diagnostic/84034
29293 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
29294 Handle CR like TAB.
29295 (layout::print_source_line): Likewise.
29296 (test_get_line_width_without_trailing_whitespace): Add test cases.
29298 2018-01-27 Jakub Jelinek <jakub@redhat.com>
29300 PR middle-end/84040
29301 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
29304 2018-01-26 Jim Wilson <jimw@sifive.com>
29306 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
29308 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
29311 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29313 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
29314 and CMP + SUB-immediate -> SUBS.
29316 2018-01-26 Martin Sebor <msebor@redhat.com>
29318 PR tree-optimization/83896
29319 * tree-ssa-strlen.c (get_string_len): Rename...
29320 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
29321 Avoid assuming length is constant.
29322 (handle_char_store): Use HOST_WIDE_INT for string length.
29324 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
29327 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
29328 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
29330 2018-01-26 Richard Biener <rguenther@suse.de>
29332 PR rtl-optimization/84003
29333 * dse.c (record_store): Only record redundant stores when
29334 the earlier store aliases at least all accesses the later one does.
29336 2018-01-26 Jakub Jelinek <jakub@redhat.com>
29338 PR rtl-optimization/83985
29339 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
29340 REG_CFA_RESTORE insns.
29341 (delete_unmarked_insns): Don't ignore separate shrink wrapping
29342 REG_CFA_RESTORE insns here.
29345 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
29346 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
29348 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
29350 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
29351 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
29352 (arc_init): Likewise.
29353 (arc_override_options): Likewise.
29354 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
29356 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
29358 * config/arc/arc.h (TARGET_DBNZ): Define.
29359 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
29360 properly set the tune attribute.
29361 (dbnz): Use TARGET_DBNZ guard.
29362 * config/arc/arc.opt (mtune): Add core3 option.
29364 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
29366 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
29367 recognize new pic like addresses.
29368 (arc_delegitimize_address): Clean up.
29370 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
29372 * config/arc/arc-arches.def: Option mrf16 valid for all
29374 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
29375 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
29376 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
29377 * config/arc/arc-tables.opt: Regenerate.
29378 * config/arc/arc.c (arc_conditional_register_usage): Handle
29379 reduced register file case.
29380 (arc_file_start): Set must have build attributes.
29381 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
29382 mrf16 option value.
29383 * config/arc/arc.opt (mrf16): Add new option.
29384 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
29385 * config/arc/genmultilib.awk: Handle new mrf16 option.
29386 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
29387 * config/arc/t-multilib: Regenerate.
29388 * doc/invoke.texi (ARC Options): Document mrf16 option.
29390 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
29392 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
29393 * config/arc/arc.c (arc_handle_secure_attribute): New function.
29394 (arc_attribute_table): Add 'secure_call' attribute.
29395 (arc_print_operand): Print secure call operand.
29396 (arc_function_ok_for_sibcall): Don't optimize tail calls when
29398 (arc_is_secure_call_p): New function. * config/arc/arc.md
29399 (call_i): Add support for sjli instruction.
29400 (call_value_i): Likewise.
29401 * config/arc/constraints.md (Csc): New constraint.
29403 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
29404 John Eric Martin <John.Martin@emmicro-us.com>
29406 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
29407 * config/arc/arc.c (_arc_jli_section): New struct.
29408 (arc_jli_section): New type.
29409 (rc_jli_sections): New static variable.
29410 (arc_handle_jli_attribute): New function.
29411 (arc_attribute_table): Add jli_always and jli_fixed attribute.
29412 (arc_file_end): New function.
29413 (TARGET_ASM_FILE_END): Define.
29414 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
29415 (arc_add_jli_section): New function.
29416 (jli_call_scan): Likewise.
29417 (arc_reorg): Call jli_call_scan.
29418 (arc_output_addsi): Remove 'S' from printing asm operand.
29419 (arc_is_jli_call_p): New function.
29420 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
29422 (movhi_insn): Likewise.
29423 (movsi_insn): Likewise.
29424 (movsi_set_cc_insn): Likewise.
29425 (loadqi_update): Likewise.
29426 (load_zeroextendqisi_update): Likewise.
29427 (load_signextendqisi_update): Likewise.
29428 (loadhi_update): Likewise.
29429 (load_zeroextendhisi_update): Likewise.
29430 (load_signextendhisi_update): Likewise.
29431 (loadsi_update): Likewise.
29432 (loadsf_update): Likewise.
29433 (movsicc_insn): Likewise.
29434 (bset_insn): Likewise.
29435 (bxor_insn): Likewise.
29436 (bclr_insn): Likewise.
29437 (bmsk_insn): Likewise.
29438 (bicsi3_insn): Likewise.
29439 (cmpsi_cc_c_insn): Likewise.
29440 (movsi_ne): Likewise.
29441 (movsi_cond_exec): Likewise.
29442 (clrsbsi2): Likewise.
29443 (norm_f): Likewise.
29453 (call_i): Remove 'S' asm letter, add jli instruction.
29454 (call_value_i): Likewise.
29455 * config/arc/arc.op (mjli-always): New option.
29456 * config/arc/constraints.md (Cji): New constraint.
29457 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
29459 (subsf3_fpx): Likewise.
29460 (mulsf3_fpx): Likewise.
29461 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
29463 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
29464 function attrbutes.
29465 * doc/invoke.texi (ARC): Document mjli-always option.
29467 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
29469 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
29470 avoid addition with 0 and use incw and decw where possible.
29472 2018-01-26 Richard Biener <rguenther@suse.de>
29474 PR tree-optimization/81082
29475 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
29476 association if it requires casting to unsigned.
29477 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
29478 from fold_plusminus_mult_expr to catch important cases late when
29479 range info is available.
29481 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29483 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
29484 * configure.ac (hidden_linkonce): New test.
29485 * configure: Regenerate.
29486 * config.in: Regenerate.
29488 2018-01-26 Julia Koval <julia.koval@intel.com>
29490 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
29491 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
29492 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
29493 _mm_mask_bitshuffle_epi64_mask): Fix type.
29494 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
29495 USI_FTYPE_V4DI_V4DI_USI): Remove.
29496 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
29497 __builtin_ia32_vpshufbitqmb256_mask,
29498 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
29499 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
29500 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
29502 2018-01-26 Alan Modra <amodra@gmail.com>
29505 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
29506 UNSPEC_VBPERMQ. Sort other unspecs.
29508 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
29510 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
29512 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
29514 PR middle-end/83055
29515 * predict.c (drop_profile): Do not push/pop cfun; update also
29517 (handle_missing_profiles): Fix logic looking for zero profiles.
29519 2018-01-25 Jakub Jelinek <jakub@redhat.com>
29521 PR middle-end/83977
29522 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
29523 on functions with #pragma omp declare simd or functions with simd
29525 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
29526 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
29527 Remove trailing \n from warning_at calls.
29529 2018-01-25 Tom de Vries <tom@codesourcery.com>
29532 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
29533 for neutered workers.
29535 2018-01-24 Joseph Myers <joseph@codesourcery.com>
29538 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
29539 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
29541 2018-01-24 Jeff Law <law@redhat.com>
29544 * i386.c (get_probe_interval): Move to earlier point.
29545 (ix86_compute_frame_layout): If -fstack-clash-protection and
29546 the frame is larger than the probe interval, then use pushes
29547 to save registers rather than reg->mem moves.
29548 (ix86_expand_prologue): Remove conditional for int_registers_saved
29551 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
29554 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
29555 min/max for never referenced object.
29557 2018-01-24 Jakub Jelinek <jakub@redhat.com>
29559 PR middle-end/83977
29560 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
29562 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
29563 attributes from DECL_ATTRIBUTES (decl) without affecting
29564 DECL_ATTRIBUTES (current_function_decl).
29565 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
29566 functions with non-NULL DECL_ABSTRACT_ORIGIN.
29568 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
29570 PR tree-optimization/83979
29571 * fold-const.c (fold_comparison): Use constant_boolean_node
29572 instead of boolean_{true,false}_node.
29574 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
29576 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
29579 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29581 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
29582 Simplify the clause that sets the length attribute.
29583 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
29584 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
29585 clause that sets the length attribute.
29586 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
29588 2018-01-24 Tom de Vries <tom@codesourcery.com>
29591 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
29592 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
29593 Add strict parameter.
29594 (prevent_branch_around_nothing): Insert dummy insn between branch to
29595 label and label with no ptx insn inbetween.
29596 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
29598 2018-01-24 Tom de Vries <tom@codesourcery.com>
29601 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
29602 for neutered threads in warp.
29603 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
29605 2018-01-24 Richard Biener <rguenther@suse.de>
29607 PR tree-optimization/83176
29608 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
29611 2018-01-24 Richard Biener <rguenther@suse.de>
29613 PR tree-optimization/82819
29614 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
29615 code generating pluses that are no-ops in the target precision.
29617 2018-01-24 Richard Biener <rguenther@suse.de>
29619 PR middle-end/84000
29620 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
29622 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
29624 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
29625 to merge probabilities.
29626 * predict.c (probably_never_executed): Also mark as cold functions
29627 with global 0 profile and guessed local profile.
29628 * profile-count.c (profile_probability::combine_with_count): New
29630 * profile-count.h (profile_probability::operator*,
29631 profile_probability::operator*=, profile_probability::operator/,
29632 profile_probability::operator/=): Reduce precision to adjusted
29633 and set value to guessed on contradictory divisions.
29634 (profile_probability::combine_with_freq): Remove.
29635 (profile_probability::combine_wiht_count): Declare.
29636 (profile_count::force_nonzero):: Set to adjusted.
29637 (profile_count::probability_in):: Set quality to adjusted.
29638 * tree-ssa-tail-merge.c (replace_block_by): Use
29639 combine_with_count.
29641 2018-01-23 Andrew Waterman <andrew@sifive.com>
29642 Jim Wilson <jimw@sifive.com>
29644 * config/riscv/riscv.c (riscv_stack_boundary): New.
29645 (riscv_option_override): Set riscv_stack_boundary. Handle
29646 riscv_preferred_stack_boundary_arg.
29647 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
29648 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
29649 (STACK_BOUNDARY): Set to riscv_stack_boundary.
29650 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
29651 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
29652 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
29654 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
29657 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
29658 of struct ix86_frame.
29659 (ix86_expand_epilogue): Likewise. Add a local variable for
29660 the reg_save_offset field in struct ix86_frame.
29662 2018-01-23 Bin Cheng <bin.cheng@arm.com>
29664 PR tree-optimization/82604
29665 * tree-loop-distribution.c (enum partition_kind): New enum item
29666 PKIND_PARTIAL_MEMSET.
29667 (partition_builtin_p): Support above new enum item.
29668 (generate_code_for_partition): Ditto.
29669 (compute_access_range): Differentiate cases that equality can be
29670 proven at all loops, the innermost loops or no loops.
29671 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
29672 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
29673 (finalize_partitions, distribute_loop): Don't fuse partition of
29674 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
29675 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
29676 parloop is enabled.
29678 2018-01-23 Martin Liska <mliska@suse.cz>
29680 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
29681 order to ignore the predictor.
29682 (PRED_POLYMORPHIC_CALL): Likewise.
29683 (PRED_RECURSIVE_CALL): Likewise.
29685 2018-01-23 Martin Liska <mliska@suse.cz>
29687 * tree-profile.c (tree_profiling): Print function header to
29688 aware reader which function we are working on.
29689 * value-prof.c (gimple_find_values_to_profile): Do not print
29690 not interesting value histograms.
29692 2018-01-23 Martin Liska <mliska@suse.cz>
29694 * profile-count.h (enum profile_quality): Add
29695 profile_uninitialized as the first value. Do not number values
29696 as they are zero based.
29697 (profile_count::verify): Update sanity check.
29698 (profile_probability::verify): Likewise.
29700 2018-01-23 Nathan Sidwell <nathan@acm.org>
29702 * doc/invoke.texi (ffor-scope): Deprecate.
29704 2018-01-23 David Malcolm <dmalcolm@redhat.com>
29706 PR tree-optimization/83510
29707 * domwalk.c (set_all_edges_as_executable): New function.
29708 (dom_walker::dom_walker): Convert bool param
29709 "skip_unreachable_blocks" to enum reachability. Move setup of
29710 edge flags to set_all_edges_as_executable and only do it when
29711 reachability is REACHABLE_BLOCKS.
29712 * domwalk.h (enum dom_walker::reachability): New enum.
29713 (dom_walker::dom_walker): Convert bool param
29714 "skip_unreachable_blocks" to enum reachability.
29715 (set_all_edges_as_executable): New decl.
29716 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
29717 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
29719 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
29720 but converting true to REACHABLE_BLOCKS.
29721 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
29723 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
29724 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
29725 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
29727 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
29728 if check_all_array_refs will be called.
29730 2018-01-23 David Malcolm <dmalcolm@redhat.com>
29732 * tree.c (selftest::test_location_wrappers): Add more test
29735 2018-01-23 David Malcolm <dmalcolm@redhat.com>
29737 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
29738 (selftest::test_bit_in_range): Likewise.
29740 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
29743 * doc/sourcebuild.texi (vect_float): Say that the selector
29744 only describes the situation when -funsafe-math-optimizations is on.
29745 (vect_float_strict): Document.
29747 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
29749 PR tree-optimization/83965
29750 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
29751 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
29752 instead of checking only for a reduction.
29753 (vect_recog_widen_sum_pattern): Likewise.
29755 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
29757 * predict.c (probably_never_executed): Only use precise profile info.
29758 (compute_function_frequency): Skip after inlining hack since we now
29759 have quality checking.
29761 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
29763 * profile-count.h (profile_probability::very_unlikely,
29764 profile_probability::unlikely, profile_probability::even): Set
29765 precision to guessed.
29767 2018-01-23 Richard Biener <rguenther@suse.de>
29769 PR tree-optimization/83963
29770 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
29771 Properly terminate dominator walk when crossing the exit edge not
29772 when visiting its source block.
29774 2018-01-23 Jakub Jelinek <jakub@redhat.com>
29777 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
29778 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
29780 2018-01-22 Jakub Jelinek <jakub@redhat.com>
29782 PR tree-optimization/83957
29783 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
29784 semicolon after for body surrounded by braces.
29786 PR tree-optimization/83081
29787 * profile-count.h (profile_probability::split): New method.
29788 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
29789 Use profile_probability::split.
29790 (do_compare_rtx_and_jump): Fix adjustment of probabilities
29791 when splitting a single conditional jump into 2.
29793 2018-01-22 David Malcolm <dmalcolm@redhat.com>
29795 PR tree-optimization/69452
29796 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
29799 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29801 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
29802 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
29803 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
29805 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29807 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
29809 * config/rl78/rl78.md (movdi): New define_expand.
29810 * config/rl78/rl78.c (rl78_split_movdi): New function.
29812 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29815 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
29817 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
29818 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
29819 128-bit to produce an UNSPEC move to get the double word with the
29820 signbit and then a shift directly to do signbit.
29821 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
29822 implementation with a new version that just does either a direct
29823 move or a regular move. Move memory interface to separate insns.
29824 Move insns so they are next to the expander.
29825 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
29826 with signbit move. Split big and little endian case.
29827 (signbit<mode>2_dm_mem_le): Likewise.
29828 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
29829 (signbit<mode>2_dm2): Likewise.
29831 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29833 * config/rl78/rl78.md (anddi3): New define_expand.
29835 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29837 * config/rl78/rl78.md (umindi3): New define_expand.
29839 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29841 * config/rl78/rl78.md (smindi3): New define_expand.
29843 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29845 * config/rl78/rl78.md (smaxdi3): New define_expand.
29847 2018-01-22 Carl Love <cel@us.ibm.com>
29849 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
29850 LVX_V1TI): Add macro expansion.
29851 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
29852 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
29853 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
29854 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
29855 Change check to determine if the instruction is a byte reversing
29856 entry. Fix typo in comment.
29857 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
29858 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
29859 Add def_builtin calls for new builtins.
29860 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
29861 Add define_insn expansion.
29863 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29865 * config/rl78/rl78.md (umaxdi3): New define_expand.
29867 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29869 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
29870 for non-QImode registers.
29872 2018-01-22 Richard Biener <rguenther@suse.de>
29874 PR tree-optimization/83963
29875 * graphite-scop-detection.c (scop_detection::get_sese): Delay
29876 including the loop exit block.
29877 (scop_detection::merge_sese): Likewise.
29878 (scop_detection::add_scop): Do it here instead.
29880 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29882 * doc/sourcebuild.texi (arm_softfloat): Document.
29884 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
29887 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
29888 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
29889 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
29891 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29892 David Edelsohn <dje.gcc@gmail.com>
29895 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
29896 Change "crset eq" to "crset 2".
29897 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
29898 (*call_indirect_aix<mode>_nospec): Likewise.
29899 (*call_value_indirect_aix<mode>_nospec): Likewise.
29900 (*call_indirect_elfv2<mode>_nospec): Likewise.
29901 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
29902 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
29903 change assembly output from . to $.
29904 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
29905 (indirect_jump<mode>_nospec): Change assembly output from . to $.
29906 (*tablejump<mode>_internal1_nospec): Likewise.
29908 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
29911 * config/sh/sh_optimize_sett_clrt.cc:
29912 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
29914 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
29916 PR tree-optimization/83940
29917 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
29918 offset_dt to vect_constant_def rather than vect_unknown_def_type.
29919 (vect_check_load_store_mask): Add a mask_dt_out parameter and
29920 use it to pass back the definition type.
29921 (vect_check_store_rhs): Likewise rhs_dt_out.
29922 (vect_build_gather_load_calls): Add a mask_dt argument and use
29923 it instead of a call to vect_is_simple_use.
29924 (vectorizable_store): Update calls to vect_check_load_store_mask
29925 and vect_check_store_rhs. Use the dt returned by the latter instead
29926 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
29927 instead of calls to vect_is_simple_use. Pass the scalar rather
29928 than the vector operand to vect_is_simple_use when handling
29929 second and subsequent copies of an rhs value.
29930 (vectorizable_load): Update calls to vect_check_load_store_mask
29931 and vect_build_gather_load_calls. Use the cached mask_dt and
29932 gs_info.offset_dt instead of calls to vect_is_simple_use.
29934 2018-01-20 Jakub Jelinek <jakub@redhat.com>
29936 PR middle-end/83945
29937 * tree-emutls.c: Include gimplify.h.
29938 (lower_emutls_2): New function.
29939 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
29940 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
29941 it before further processing.
29944 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
29945 UINTVAL (trueop1) instead of INTVAL (op1).
29947 2018-01-19 Jakub Jelinek <jakub@redhat.com>
29951 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
29952 INCOMING_FRAME_SP_OFFSET if not defined.
29953 (scan_trace): Add ENTRY argument. If true and
29954 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
29955 emit a note to adjust the CFA offset.
29956 (create_cfi_notes): Adjust scan_trace callers.
29957 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
29958 INCOMING_FRAME_SP_OFFSET in the CIE.
29959 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
29960 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
29962 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
29963 * doc/tm.texi: Regenerated.
29965 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29967 PR rtl-optimization/83147
29968 * lra-constraints.c (remove_inheritance_pseudos): Use
29969 lra_substitute_pseudo_within_insn.
29971 2018-01-19 Tom de Vries <tom@codesourcery.com>
29972 Cesar Philippidis <cesar@codesourcery.com>
29975 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
29977 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
29980 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
29981 spaces for function labels.
29983 2018-01-19 Martin Liska <mliska@suse.cz>
29985 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
29986 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
29987 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
29988 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
29989 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
29990 (PRED_CONST_RETURN): Change from 69 to 65.
29991 (PRED_NULL_RETURN): Change from 91 to 71.
29992 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
29993 (PRED_LOOP_GUARD): Change from 66 to 73.
29995 2018-01-19 Martin Liska <mliska@suse.cz>
29997 * predict.c (predict_insn_def): Add new assert.
29998 (struct branch_predictor): Change type to signed integer.
29999 (test_prediction_value_range): Amend test to cover
30000 PROB_UNINITIALIZED.
30001 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
30002 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
30003 (PRED_LOOP_ITERATIONS_MAX): Likewise.
30004 (PRED_LOOP_IV_COMPARE): Likewise.
30005 * predict.h (PROB_UNINITIALIZED): Define new constant.
30007 2018-01-19 Martin Liska <mliska@suse.cz>
30009 * predict.c (dump_prediction): Add new format for
30010 analyze_brprob.py script which is enabled with -details
30012 * profile-count.h (precise_p): New function.
30014 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
30016 PR tree-optimization/83922
30017 * tree-vect-loop.c (vect_verify_full_masking): Return false if
30018 there are no statements that need masking.
30019 (vect_active_double_reduction_p): New function.
30020 (vect_analyze_loop_operations): Use it when handling phis that
30021 are not in the loop header.
30023 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
30025 PR tree-optimization/83914
30026 * tree-vect-loop.c (vectorizable_induction): Don't convert
30027 init_expr or apply the peeling adjustment for inductions
30028 that are nested within the vectorized loop.
30030 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30032 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
30035 2018-01-18 Jakub Jelinek <jakub@redhat.com>
30039 * function.h (gimplify_parameters): Add gimple_seq * argument.
30040 * function.c: Include gimple.h and options.h.
30041 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
30042 for the added local temporaries if needed.
30043 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
30044 if there are any parameter cleanups, wrap whole body into a
30045 try/finally with the cleanups.
30047 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
30050 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
30051 Use GET_MODE_CLASS for scalar floating point.
30053 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
30057 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
30058 Fix call of call_cgraph_insertion_hooks.
30060 2018-01-18 Martin Sebor <msebor@redhat.com>
30062 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
30064 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
30067 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
30070 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
30073 * common.opt: (-ffile-prefix-map): New option.
30074 * opts.c (common_handle_option): Defer it.
30075 * opts-global.c (handle_common_deferred_options): Handle it.
30076 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
30077 * file-prefix-map.h: New file.
30078 (remap_debug_filename, add_debug_prefix_map): ...here.
30079 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
30080 * final.c (debug_prefix_map, add_debug_prefix_map
30081 remap_debug_filename): Move to...
30082 * file-prefix-map.c: New file.
30083 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
30084 generalize, get rid of alloca(), use strrchr() instead of strchr().
30085 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
30086 Implement in terms of add_prefix_map().
30087 (remap_macro_filename, remap_debug_filename): Implement in term of
30089 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
30090 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
30091 * dbxout.c: Include file-prefix-map.h.
30092 * varasm.c: Likewise.
30093 * vmsdbgout.c: Likewise.
30094 * xcoffout.c: Likewise.
30095 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
30096 * doc/cppopts.texi (-fmacro-prefix-map): Document.
30097 * doc/invoke.texi (-ffile-prefix-map): Document.
30098 (-fdebug-prefix-map): Update description.
30100 2018-01-18 Martin Liska <mliska@suse.cz>
30102 * config/i386/i386.c (indirect_thunk_name): Document that also
30104 (output_indirect_thunk): Document why both instructions
30105 (pause and lfence) are generated.
30107 2018-01-18 Richard Biener <rguenther@suse.de>
30109 PR tree-optimization/83887
30110 * graphite-scop-detection.c
30111 (scop_detection::get_nearest_dom_with_single_entry): Remove.
30112 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
30113 (scop_detection::merge_sese): Re-implement with a flood-fill
30114 algorithm that properly finds a SESE region if it exists.
30116 2018-01-18 Jakub Jelinek <jakub@redhat.com>
30119 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
30120 pointer_diff optimizations use view_convert instead of convert.
30122 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30124 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
30125 Generate different code for -mno-speculate-indirect-jumps.
30126 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
30127 (*call_indirect_aix<mode>): Disable for
30128 -mno-speculate-indirect-jumps.
30129 (*call_indirect_aix<mode>_nospec): New define_insn.
30130 (*call_value_indirect_aix<mode>): Disable for
30131 -mno-speculate-indirect-jumps.
30132 (*call_value_indirect_aix<mode>_nospec): New define_insn.
30133 (*sibcall_nonlocal_sysv<mode>): Generate different code for
30134 -mno-speculate-indirect-jumps.
30135 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
30137 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
30139 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
30140 long double type, set the flags for noting the default long double
30141 type, even if we don't pass or return a long double type.
30143 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
30146 * ipa-inline.c (flatten_function): Do not overwrite final inlining
30149 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
30151 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
30152 support for merge[hl].
30153 (fold_mergehl_helper): New helper function.
30154 (tree-vector-builder.h): New #include for tree_vector_builder usage.
30155 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
30156 (altivec_vmrglw_direct): Add xxmrglw insn.
30158 2018-01-17 Andrew Waterman <andrew@sifive.com>
30160 * config/riscv/riscv.c (riscv_conditional_register_usage): If
30161 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
30163 2018-01-17 David Malcolm <dmalcolm@redhat.com>
30166 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
30167 call the lto_location_cache before reading the
30168 DECL_SOURCE_LOCATION of the types.
30170 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
30171 Richard Sandiford <richard.sandiford@linaro.org>
30173 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
30174 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
30175 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
30176 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
30177 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
30179 * config/aarch64/constraints.md (aarch64_movti_operand):
30181 * config/aarch64/predicates.md (Uti): Add new constraint.
30183 2018-01-17 Carl Love <cel@us.ibm.com>
30185 * config/rs6000/vsx.md (define_expand xl_len_r,
30186 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
30187 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
30189 (define_expand, define_insn): Move the shift left from the
30190 define_insn to the define_expand for lxvl and stxvl instructions.
30191 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
30192 and XL_LEN_R definitions to PURE.
30194 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
30196 * config/i386/i386.c (indirect_thunk_name): Declare regno
30197 as unsigned int. Compare regno with INVALID_REGNUM.
30198 (output_indirect_thunk): Ditto.
30199 (output_indirect_thunk_function): Ditto.
30200 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
30201 in the call to output_indirect_thunk_function.
30203 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
30205 PR middle-end/83884
30206 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
30207 rather than the size of inner_type to determine the stack slot size
30208 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
30210 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
30213 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
30216 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
30218 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
30219 endian Linux systems to optionally enable multilibs for selecting
30220 the long double type if the user configured an explicit type.
30221 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
30222 have no long double multilibs if not defined.
30223 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
30224 warn if the user used -mabi={ieee,ibm}longdouble and we built
30225 multilibs for long double.
30226 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
30227 appropriate multilib option.
30228 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
30230 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
30231 for building long double multilibs.
30232 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
30234 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
30236 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
30239 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
30241 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
30244 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
30247 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
30250 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
30252 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
30253 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
30255 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
30257 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
30258 different rtl trees depending on TARGET_64BIT.
30259 (rs6000_gen_lvx): Likewise.
30261 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
30263 * config/visium/visium.md (nop): Tweak comment.
30264 (hazard_nop): Likewise.
30266 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30268 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
30269 -mspeculate-indirect-jumps.
30270 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
30271 for -mno-speculate-indirect-jumps.
30272 (*call_indirect_elfv2<mode>_nospec): New define_insn.
30273 (*call_value_indirect_elfv2<mode>): Disable for
30274 -mno-speculate-indirect-jumps.
30275 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
30276 (indirect_jump): Emit different RTL for
30277 -mno-speculate-indirect-jumps.
30278 (*indirect_jump<mode>): Disable for
30279 -mno-speculate-indirect-jumps.
30280 (*indirect_jump<mode>_nospec): New define_insn.
30281 (tablejump): Emit different RTL for
30282 -mno-speculate-indirect-jumps.
30283 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
30284 (tablejumpsi_nospec): New define_expand.
30285 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
30286 (tablejumpdi_nospec): New define_expand.
30287 (*tablejump<mode>_internal1): Disable for
30288 -mno-speculate-indirect-jumps.
30289 (*tablejump<mode>_internal1_nospec): New define_insn.
30290 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
30293 2018-01-16 Artyom Skrobov tyomitch@gmail.com
30295 * caller-save.c (insert_save): Drop unnecessary parameter. All
30298 2018-01-16 Jakub Jelinek <jakub@redhat.com>
30299 Richard Biener <rguenth@suse.de>
30302 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
30303 return early, inline manually is_gimple_sizepos. Make sure if we
30304 call gimplify_expr we don't end up with a gimple constant.
30305 * tree.c (variably_modified_type_p): Don't return true for
30306 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
30307 * gimplify.h (is_gimple_sizepos): Remove.
30309 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
30311 PR tree-optimization/83857
30312 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
30313 vectorizable_live_operation for pure SLP statements.
30314 (vectorizable_live_operation): Handle PHIs.
30316 2018-01-16 Richard Biener <rguenther@suse.de>
30318 PR tree-optimization/83867
30319 * tree-vect-stmts.c (vect_transform_stmt): Precompute
30320 nested_in_vect_loop_p since the scalar stmt may get invalidated.
30322 2018-01-16 Jakub Jelinek <jakub@redhat.com>
30325 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
30326 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
30327 If off is not INTEGER_CST, issue a may not be aligned warning
30328 rather than isn't aligned. Use isn%'t rather than isn't.
30329 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
30331 <case MULT_EXPR>: Improve the case when bottom and one of the
30332 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
30333 operand, in that case check if the other operand is multiple of
30334 bottom divided by the INTEGER_CST operand.
30336 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
30339 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
30340 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
30341 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
30342 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
30343 * config/pa/pa.c (pa_function_arg_advance): Likewise.
30344 (pa_function_arg, pa_arg_partial_bytes): Likewise.
30345 (pa_function_arg_size): New function.
30347 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
30349 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
30350 in a separate statement.
30352 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
30354 PR tree-optimization/83847
30355 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
30356 group gathers and scatters.
30358 2018-01-16 Jakub Jelinek <jakub@redhat.com>
30360 PR rtl-optimization/86620
30361 * params.def (max-sched-ready-insns): Bump minimum value to 1.
30363 PR rtl-optimization/83213
30364 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
30365 to last if both are JUMP_INSNs.
30367 PR tree-optimization/83843
30368 * gimple-ssa-store-merging.c
30369 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
30370 store_immediate_info for bswap/nop orig_stores.
30372 2018-01-15 Andrew Waterman <andrew@sifive.com>
30374 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
30376 <UDIV>: Increase cost if !TARGET_DIV.
30378 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30380 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
30381 (define_attr "cr_logical_3op"): New.
30382 (cceq_ior_compare): Adjust.
30383 (cceq_ior_compare_complement): Adjust.
30384 (*cceq_rev_compare): Adjust.
30385 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
30386 (is_cracked_insn): Adjust.
30387 (insn_must_be_first_in_group): Adjust.
30388 * config/rs6000/40x.md: Adjust.
30389 * config/rs6000/440.md: Adjust.
30390 * config/rs6000/476.md: Adjust.
30391 * config/rs6000/601.md: Adjust.
30392 * config/rs6000/603.md: Adjust.
30393 * config/rs6000/6xx.md: Adjust.
30394 * config/rs6000/7450.md: Adjust.
30395 * config/rs6000/7xx.md: Adjust.
30396 * config/rs6000/8540.md: Adjust.
30397 * config/rs6000/cell.md: Adjust.
30398 * config/rs6000/e300c2c3.md: Adjust.
30399 * config/rs6000/e500mc.md: Adjust.
30400 * config/rs6000/e500mc64.md: Adjust.
30401 * config/rs6000/e5500.md: Adjust.
30402 * config/rs6000/e6500.md: Adjust.
30403 * config/rs6000/mpc.md: Adjust.
30404 * config/rs6000/power4.md: Adjust.
30405 * config/rs6000/power5.md: Adjust.
30406 * config/rs6000/power6.md: Adjust.
30407 * config/rs6000/power7.md: Adjust.
30408 * config/rs6000/power8.md: Adjust.
30409 * config/rs6000/power9.md: Adjust.
30410 * config/rs6000/rs64.md: Adjust.
30411 * config/rs6000/titan.md: Adjust.
30413 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
30415 * config/i386/predicates.md (indirect_branch_operand): Rewrite
30416 ix86_indirect_branch_register logic.
30418 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
30420 * config/i386/constraints.md (Bs): Update
30421 ix86_indirect_branch_register check. Don't check
30422 ix86_indirect_branch_register with GOT_memory_operand.
30424 * config/i386/predicates.md (GOT_memory_operand): Don't check
30425 ix86_indirect_branch_register here.
30426 (GOT32_symbol_operand): Likewise.
30428 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
30430 * config/i386/predicates.md (constant_call_address_operand):
30431 Rewrite ix86_indirect_branch_register logic.
30432 (sibcall_insn_operand): Likewise.
30434 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
30436 * config/i386/constraints.md (Bs): Replace
30437 ix86_indirect_branch_thunk_register with
30438 ix86_indirect_branch_register.
30440 * config/i386/i386.md (indirect_jump): Likewise.
30441 (tablejump): Likewise.
30442 (*sibcall_memory): Likewise.
30443 (*sibcall_value_memory): Likewise.
30444 Peepholes of indirect call and jump via memory: Likewise.
30445 * config/i386/i386.opt: Likewise.
30446 * config/i386/predicates.md (indirect_branch_operand): Likewise.
30447 (GOT_memory_operand): Likewise.
30448 (call_insn_operand): Likewise.
30449 (sibcall_insn_operand): Likewise.
30450 (GOT32_symbol_operand): Likewise.
30452 2018-01-15 Jakub Jelinek <jakub@redhat.com>
30454 PR middle-end/83837
30455 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
30456 type rather than type addr's type points to.
30457 (expand_omp_atomic_mutex): Likewise.
30458 (expand_omp_atomic): Likewise.
30460 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
30463 * config/i386/i386.c (output_indirect_thunk_function): Use
30464 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
30465 for __x86_return_thunk.
30467 2018-01-15 Richard Biener <rguenther@suse.de>
30469 PR middle-end/83850
30470 * expmed.c (extract_bit_field_1): Fix typo.
30472 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30475 * config/arm/iterators.md (VF): New mode iterator.
30476 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
30477 Remove integer-related logic from pattern.
30478 (neon_vabd<mode>_3): Likewise.
30480 2018-01-15 Jakub Jelinek <jakub@redhat.com>
30482 PR middle-end/82694
30483 * common.opt (fstrict-overflow): No longer an alias.
30484 (fwrapv-pointer): New option.
30485 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
30486 also for pointer types based on flag_wrapv_pointer.
30487 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
30488 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
30489 opts->x_flag_wrapv got set.
30490 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
30491 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
30492 POINTER_TYPE_OVERFLOW_UNDEFINED.
30493 * match.pd: Likewise in address comparison pattern.
30494 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
30496 2018-01-15 Richard Biener <rguenther@suse.de>
30499 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
30500 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
30501 Reset type names to their identifier if their TYPE_DECL doesn't
30502 have linkage (and thus is used for ODR and devirt).
30503 (save_debug_info_for_decl): Remove.
30504 (save_debug_info_for_type): Likewise.
30505 (add_tree_to_fld_list): Adjust.
30506 * tree-pretty-print.c (dump_generic_node): Make dumping of
30507 type names more robust.
30509 2018-01-15 Richard Biener <rguenther@suse.de>
30511 * BASE-VER: Bump to 8.0.1.
30513 2018-01-14 Martin Sebor <msebor@redhat.com>
30516 * builtins.c (check_access): Avoid warning when the no-warning bit
30519 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
30521 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
30522 * ira-color (allocno_hard_regs_compare): Likewise.
30524 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
30527 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
30528 Use .pushsection/.popsection.
30530 2018-01-14 Martin Sebor <msebor@redhat.com>
30533 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
30535 2018-01-14 Jakub Jelinek <jakub@redhat.com>
30537 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
30538 entry from extra_headers.
30539 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
30540 extra_headers, make the list bitwise identical to the i?86-*-* one.
30542 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
30544 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
30545 -mcmodel=large with -mindirect-branch=thunk,
30546 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
30547 -mfunction-return=thunk-extern.
30548 * doc/invoke.texi: Document -mcmodel=large is incompatible with
30549 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
30550 -mfunction-return=thunk and -mfunction-return=thunk-extern.
30552 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
30554 * config/i386/i386.c (print_reg): Print the name of the full
30555 integer register without '%'.
30556 (ix86_print_operand): Handle 'V'.
30557 * doc/extend.texi: Document 'V' modifier.
30559 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
30561 * config/i386/constraints.md (Bs): Disallow memory operand for
30562 -mindirect-branch-register.
30564 * config/i386/predicates.md (indirect_branch_operand): Likewise.
30565 (GOT_memory_operand): Likewise.
30566 (call_insn_operand): Likewise.
30567 (sibcall_insn_operand): Likewise.
30568 (GOT32_symbol_operand): Likewise.
30569 * config/i386/i386.md (indirect_jump): Call convert_memory_address
30570 for -mindirect-branch-register.
30571 (tablejump): Likewise.
30572 (*sibcall_memory): Likewise.
30573 (*sibcall_value_memory): Likewise.
30574 Disallow peepholes of indirect call and jump via memory for
30575 -mindirect-branch-register.
30576 (*call_pop): Replace m with Bw.
30577 (*call_value_pop): Likewise.
30578 (*sibcall_pop_memory): Replace m with Bs.
30579 * config/i386/i386.opt (mindirect-branch-register): New option.
30580 * doc/invoke.texi: Document -mindirect-branch-register option.
30582 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
30584 * config/i386/i386-protos.h (ix86_output_function_return): New.
30585 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
30586 set function_return_type.
30587 (indirect_thunk_name): Add ret_p to indicate thunk for function
30589 (output_indirect_thunk_function): Pass false to
30590 indirect_thunk_name.
30591 (ix86_output_indirect_branch_via_reg): Likewise.
30592 (ix86_output_indirect_branch_via_push): Likewise.
30593 (output_indirect_thunk_function): Create alias for function
30594 return thunk if regno < 0.
30595 (ix86_output_function_return): New function.
30596 (ix86_handle_fndecl_attribute): Handle function_return.
30597 (ix86_attribute_table): Add function_return.
30598 * config/i386/i386.h (machine_function): Add
30599 function_return_type.
30600 * config/i386/i386.md (simple_return_internal): Use
30601 ix86_output_function_return.
30602 (simple_return_internal_long): Likewise.
30603 * config/i386/i386.opt (mfunction-return=): New option.
30604 (indirect_branch): Mention -mfunction-return=.
30605 * doc/extend.texi: Document function_return function attribute.
30606 * doc/invoke.texi: Document -mfunction-return= option.
30608 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
30610 * config/i386/i386-opts.h (indirect_branch): New.
30611 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
30612 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
30613 with local indirect jump when converting indirect call and jump.
30614 (ix86_set_indirect_branch_type): New.
30615 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
30616 (indirectlabelno): New.
30617 (indirect_thunk_needed): Likewise.
30618 (indirect_thunk_bnd_needed): Likewise.
30619 (indirect_thunks_used): Likewise.
30620 (indirect_thunks_bnd_used): Likewise.
30621 (INDIRECT_LABEL): Likewise.
30622 (indirect_thunk_name): Likewise.
30623 (output_indirect_thunk): Likewise.
30624 (output_indirect_thunk_function): Likewise.
30625 (ix86_output_indirect_branch_via_reg): Likewise.
30626 (ix86_output_indirect_branch_via_push): Likewise.
30627 (ix86_output_indirect_branch): Likewise.
30628 (ix86_output_indirect_jmp): Likewise.
30629 (ix86_code_end): Call output_indirect_thunk_function if needed.
30630 (ix86_output_call_insn): Call ix86_output_indirect_branch if
30632 (ix86_handle_fndecl_attribute): Handle indirect_branch.
30633 (ix86_attribute_table): Add indirect_branch.
30634 * config/i386/i386.h (machine_function): Add indirect_branch_type
30635 and has_local_indirect_jump.
30636 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
30638 (tablejump): Likewise.
30639 (*indirect_jump): Use ix86_output_indirect_jmp.
30640 (*tablejump_1): Likewise.
30641 (simple_return_indirect_internal): Likewise.
30642 * config/i386/i386.opt (mindirect-branch=): New option.
30643 (indirect_branch): New.
30646 (thunk-inline): Likewise.
30647 (thunk-extern): Likewise.
30648 * doc/extend.texi: Document indirect_branch function attribute.
30649 * doc/invoke.texi: Document -mindirect-branch= option.
30651 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
30654 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
30656 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
30658 * ipa-inline.c (want_inline_small_function_p): Return false if
30659 inlining has already failed with CIF_FINAL_ERROR.
30660 (update_caller_keys): Call want_inline_small_function_p before
30662 (update_callee_keys): Likewise.
30664 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30666 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
30668 (rs6000_quadword_masked_address_p): Likewise.
30669 (quad_aligned_load_p): Likewise.
30670 (quad_aligned_store_p): Likewise.
30671 (const_load_sequence_p): Add comment to describe the outer-most loop.
30672 (mimic_memory_attributes_and_flags): New function.
30673 (rs6000_gen_stvx): Likewise.
30674 (replace_swapped_aligned_store): Likewise.
30675 (rs6000_gen_lvx): Likewise.
30676 (replace_swapped_aligned_load): Likewise.
30677 (replace_swapped_load_constant): Capitalize argument name in
30678 comment describing this function.
30679 (rs6000_analyze_swaps): Add a third pass to search for vector loads
30680 and stores that access quad-word aligned addresses and replace
30681 with stvx or lvx instructions when appropriate.
30682 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
30683 New function prototype.
30684 (rs6000_quadword_masked_address_p): Likewise.
30685 (rs6000_gen_lvx): Likewise.
30686 (rs6000_gen_stvx): Likewise.
30687 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
30688 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
30689 when memory address is aligned.
30690 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
30691 this split to select lvx instruction when memory address is aligned.
30692 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
30693 instruction when memory address is aligned.
30694 (*vsx_le_perm_load_v16qi): Likewise.
30695 (four unnamed splitters): Modify to select the stvx instruction
30696 when memory is aligned.
30698 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
30700 * predict.c (determine_unlikely_bbs): Handle correctly BBs
30701 which appears in the queue multiple times.
30703 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30704 Alan Hayward <alan.hayward@arm.com>
30705 David Sherwood <david.sherwood@arm.com>
30707 * tree-vectorizer.h (vec_lower_bound): New structure.
30708 (_loop_vec_info): Add check_nonzero and lower_bounds.
30709 (LOOP_VINFO_CHECK_NONZERO): New macro.
30710 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
30711 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
30712 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
30713 fields. Make seg_len the distance travelled, not including the
30715 (dr_direction_indicator): Declare.
30716 (dr_zero_step_indicator): Likewise.
30717 (dr_known_forward_stride_p): Likewise.
30718 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
30720 (runtime_alias_check_p): Allow runtime alias checks with
30722 (operator ==): Compare access_size and align.
30723 (prune_runtime_alias_test_list): Rework for new distinction between
30724 the access_size and seg_len.
30725 (create_intersect_range_checks_index): Likewise. Cope with polynomial
30727 (get_segment_min_max): New function.
30728 (create_intersect_range_checks): Use it.
30729 (dr_step_indicator): New function.
30730 (dr_direction_indicator): Likewise.
30731 (dr_zero_step_indicator): Likewise.
30732 (dr_known_forward_stride_p): Likewise.
30733 * tree-loop-distribution.c (data_ref_segment_size): Return
30734 DR_STEP * (niters - 1).
30735 (compute_alias_check_pairs): Update call to the dr_with_seg_len
30737 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
30738 (vect_preserves_scalar_order_p): New function, split out from...
30739 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
30740 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
30741 (vect_vfa_access_size): New function.
30742 (vect_vfa_align): Likewise.
30743 (vect_compile_time_alias): Take access_size_a and access_b arguments.
30744 (dump_lower_bound): New function.
30745 (vect_check_lower_bound): Likewise.
30746 (vect_small_gap_p): Likewise.
30747 (vectorizable_with_step_bound_p): Likewise.
30748 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
30749 depencies if the vectorization factor is 1. Convert the checks
30750 for nonzero steps into checks on the bounds of DR_STEP. Try using
30751 a bunds check for variable steps if the minimum required step is
30752 relatively small. Update calls to the dr_with_seg_len
30753 constructor and to vect_compile_time_alias.
30754 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
30756 (vect_loop_versioning): Call it.
30757 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
30759 (vect_estimate_min_profitable_iters): Account for any bounds checks.
30761 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30762 Alan Hayward <alan.hayward@arm.com>
30763 David Sherwood <david.sherwood@arm.com>
30765 * doc/sourcebuild.texi (vect_scatter_store): Document.
30766 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
30768 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
30770 * genopinit.c (main): Add supports_vec_scatter_store and
30771 supports_vec_scatter_store_cached to target_optabs.
30772 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
30773 IFN_MASK_SCATTER_STORE.
30774 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
30776 * internal-fn.h (internal_store_fn_p): Declare.
30777 (internal_fn_stored_value_index): Likewise.
30778 * internal-fn.c (scatter_store_direct): New macro.
30779 (expand_scatter_store_optab_fn): New function.
30780 (direct_scatter_store_optab_supported_p): New macro.
30781 (internal_store_fn_p): New function.
30782 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
30783 IFN_MASK_SCATTER_STORE.
30784 (internal_fn_mask_index): Likewise.
30785 (internal_fn_stored_value_index): New function.
30786 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
30787 for scatter stores.
30788 * optabs-query.h (supports_vec_scatter_store_p): Declare.
30789 * optabs-query.c (supports_vec_scatter_store_p): New function.
30790 * tree-vectorizer.h (vect_get_store_rhs): Declare.
30791 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
30792 true for scatter stores.
30793 (vect_gather_scatter_fn_p): Handle scatter stores too.
30794 (vect_check_gather_scatter): Consider using scatter stores if
30795 supports_vec_scatter_store_p.
30796 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
30797 scatter stores too.
30798 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
30799 internal_fn_stored_value_index.
30800 (check_load_store_masking): Handle scatter stores too.
30801 (vect_get_store_rhs): Make public.
30802 (vectorizable_call): Use internal_store_fn_p.
30803 (vectorizable_store): Handle scatter store internal functions.
30804 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
30805 when deciding whether the end of the group has been reached.
30806 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
30807 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
30808 (mask_scatter_store<mode>): New insns.
30810 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30811 Alan Hayward <alan.hayward@arm.com>
30812 David Sherwood <david.sherwood@arm.com>
30814 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
30815 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
30816 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
30818 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
30819 Use vect_truncate_gather_scatter_offset if we can't treat the
30820 operation as a normal gather load or scatter store.
30821 (get_group_load_store_type): Take the gather_scatter_info
30822 as argument. Try using a gather load or scatter store for
30823 single-element groups.
30824 (get_load_store_type): Update calls to get_group_load_store_type
30825 and vect_use_strided_gather_scatters_p.
30827 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30828 Alan Hayward <alan.hayward@arm.com>
30829 David Sherwood <david.sherwood@arm.com>
30831 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
30832 optional tree argument.
30833 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
30835 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
30836 but continue to use the current value as a fallback.
30837 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
30838 to compare the updates.
30839 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
30840 (get_load_store_type): Use it when handling a strided access.
30841 (vect_get_strided_load_store_ops): New function.
30842 (vect_get_data_ptr_increment): Likewise.
30843 (vectorizable_load): Handle strided gather loads. Always pass
30844 a step to vect_create_data_ref_ptr and bump_vector_ptr.
30846 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30847 Alan Hayward <alan.hayward@arm.com>
30848 David Sherwood <david.sherwood@arm.com>
30850 * doc/md.texi (gather_load@var{m}): Document.
30851 (mask_gather_load@var{m}): Likewise.
30852 * genopinit.c (main): Add supports_vec_gather_load and
30853 supports_vec_gather_load_cached to target_optabs.
30854 * optabs-tree.c (init_tree_optimization_optabs): Use
30855 ggc_cleared_alloc to allocate target_optabs.
30856 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
30857 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
30859 * internal-fn.h (internal_load_fn_p): Declare.
30860 (internal_gather_scatter_fn_p): Likewise.
30861 (internal_fn_mask_index): Likewise.
30862 (internal_gather_scatter_fn_supported_p): Likewise.
30863 * internal-fn.c (gather_load_direct): New macro.
30864 (expand_gather_load_optab_fn): New function.
30865 (direct_gather_load_optab_supported_p): New macro.
30866 (direct_internal_fn_optab): New function.
30867 (internal_load_fn_p): Likewise.
30868 (internal_gather_scatter_fn_p): Likewise.
30869 (internal_fn_mask_index): Likewise.
30870 (internal_gather_scatter_fn_supported_p): Likewise.
30871 * optabs-query.c (supports_at_least_one_mode_p): New function.
30872 (supports_vec_gather_load_p): Likewise.
30873 * optabs-query.h (supports_vec_gather_load_p): Declare.
30874 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
30875 and memory_type field.
30876 (NUM_PATTERNS): Bump to 15.
30877 * tree-vect-data-refs.c: Include internal-fn.h.
30878 (vect_gather_scatter_fn_p): New function.
30879 (vect_describe_gather_scatter_call): Likewise.
30880 (vect_check_gather_scatter): Try using internal functions for
30881 gather loads. Recognize existing calls to a gather load function.
30882 (vect_analyze_data_refs): Consider using gather loads if
30883 supports_vec_gather_load_p.
30884 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
30885 (vect_get_gather_scatter_offset_type): Likewise.
30886 (vect_convert_mask_for_vectype): Likewise.
30887 (vect_add_conversion_to_patterm): Likewise.
30888 (vect_try_gather_scatter_pattern): Likewise.
30889 (vect_recog_gather_scatter_pattern): New pattern recognizer.
30890 (vect_vect_recog_func_ptrs): Add it.
30891 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
30892 internal_fn_mask_index and internal_gather_scatter_fn_p.
30893 (check_load_store_masking): Take the gather_scatter_info as an
30894 argument and handle gather loads.
30895 (vect_get_gather_scatter_ops): New function.
30896 (vectorizable_call): Check internal_load_fn_p.
30897 (vectorizable_load): Likewise. Handle gather load internal
30899 (vectorizable_store): Update call to check_load_store_masking.
30900 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
30901 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
30902 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
30903 (aarch64_gather_scale_operand_d): New predicates.
30904 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
30905 (mask_gather_load<mode>): New insns.
30907 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30908 Alan Hayward <alan.hayward@arm.com>
30909 David Sherwood <david.sherwood@arm.com>
30911 * optabs.def (fold_left_plus_optab): New optab.
30912 * doc/md.texi (fold_left_plus_@var{m}): Document.
30913 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
30914 * internal-fn.c (fold_left_direct): Define.
30915 (expand_fold_left_optab_fn): Likewise.
30916 (direct_fold_left_optab_supported_p): Likewise.
30917 * fold-const-call.c (fold_const_fold_left): New function.
30918 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
30919 * tree-parloops.c (valid_reduction_p): New function.
30920 (gather_scalar_reductions): Use it.
30921 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
30922 (vect_finish_replace_stmt): Declare.
30923 * tree-vect-loop.c (fold_left_reduction_fn): New function.
30924 (needs_fold_left_reduction_p): New function, split out from...
30925 (vect_is_simple_reduction): ...here. Accept reductions that
30926 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
30927 (vect_force_simple_reduction): Also store the reduction type in
30928 the assignment's STMT_VINFO_REDUC_TYPE.
30929 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
30930 (merge_with_identity): New function.
30931 (vect_expand_fold_left): Likewise.
30932 (vectorize_fold_left_reduction): Likewise.
30933 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
30934 scalar phi in place for it. Check for target support and reject
30935 cases that would reassociate the operation. Defer the transform
30936 phase to vectorize_fold_left_reduction.
30937 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
30938 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
30939 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
30941 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30943 * tree-if-conv.c (predicate_mem_writes): Remove redundant
30944 call to ifc_temp_var.
30946 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30947 Alan Hayward <alan.hayward@arm.com>
30948 David Sherwood <david.sherwood@arm.com>
30950 * target.def (legitimize_address_displacement): Take the original
30951 offset as a poly_int.
30952 * targhooks.h (default_legitimize_address_displacement): Update
30954 * targhooks.c (default_legitimize_address_displacement): Likewise.
30955 * doc/tm.texi: Regenerate.
30956 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
30957 as an argument, moving assert of ad->disp == ad->disp_term to...
30958 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
30959 Try calling targetm.legitimize_address_displacement before expanding
30960 the address rather than afterwards, and adjust for the new interface.
30961 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
30962 Match the new hook interface. Handle SVE addresses.
30963 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
30964 new hook interface.
30966 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30968 * Makefile.in (OBJS): Add early-remat.o.
30969 * target.def (select_early_remat_modes): New hook.
30970 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
30971 * doc/tm.texi: Regenerate.
30972 * targhooks.h (default_select_early_remat_modes): Declare.
30973 * targhooks.c (default_select_early_remat_modes): New function.
30974 * timevar.def (TV_EARLY_REMAT): New timevar.
30975 * passes.def (pass_early_remat): New pass.
30976 * tree-pass.h (make_pass_early_remat): Declare.
30977 * early-remat.c: New file.
30978 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
30980 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
30982 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30983 Alan Hayward <alan.hayward@arm.com>
30984 David Sherwood <david.sherwood@arm.com>
30986 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
30987 vfm1 with a bound_epilog parameter.
30988 (vect_do_peeling): Update calls accordingly, and move the prologue
30989 call earlier in the function. Treat the base bound_epilog as 0 for
30990 fully-masked loops and retain vf - 1 for other loops. Add 1 to
30991 this base when peeling for gaps.
30992 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
30993 with fully-masked loops.
30994 (vect_estimate_min_profitable_iters): Handle the single peeled
30995 iteration in that case.
30997 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30998 Alan Hayward <alan.hayward@arm.com>
30999 David Sherwood <david.sherwood@arm.com>
31001 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
31002 single-element interleaving even if the size is not a power of 2.
31003 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
31004 accesses for single-element interleaving if the group size is
31007 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31008 Alan Hayward <alan.hayward@arm.com>
31009 David Sherwood <david.sherwood@arm.com>
31011 * doc/md.texi (fold_extract_last_@var{m}): Document.
31012 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
31013 * optabs.def (fold_extract_last_optab): New optab.
31014 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
31015 * internal-fn.c (fold_extract_direct): New macro.
31016 (expand_fold_extract_optab_fn): Likewise.
31017 (direct_fold_extract_optab_supported_p): Likewise.
31018 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
31019 * tree-vect-loop.c (vect_model_reduction_cost): Handle
31020 EXTRACT_LAST_REDUCTION.
31021 (get_initial_def_for_reduction): Do not create an initial vector
31022 for EXTRACT_LAST_REDUCTION reductions.
31023 (vectorizable_reduction): Leave the scalar phi in place for
31024 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
31025 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
31026 epilogue code for EXTRACT_LAST_REDUCTION and defer the
31027 transform phase to vectorizable_condition.
31028 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
31030 (vect_finish_stmt_generation): ...here.
31031 (vect_finish_replace_stmt): New function.
31032 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
31033 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
31035 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
31037 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31038 Alan Hayward <alan.hayward@arm.com>
31039 David Sherwood <david.sherwood@arm.com>
31041 * doc/md.texi (extract_last_@var{m}): Document.
31042 * optabs.def (extract_last_optab): New optab.
31043 * internal-fn.def (EXTRACT_LAST): New internal function.
31044 * internal-fn.c (cond_unary_direct): New macro.
31045 (expand_cond_unary_optab_fn): Likewise.
31046 (direct_cond_unary_optab_supported_p): Likewise.
31047 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
31048 loops using EXTRACT_LAST.
31049 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
31050 (extract_last_<mode>): ...this optab.
31051 (vec_extract<mode><Vel>): Update accordingly.
31053 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31054 Alan Hayward <alan.hayward@arm.com>
31055 David Sherwood <david.sherwood@arm.com>
31057 * target.def (empty_mask_is_expensive): New hook.
31058 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
31059 * doc/tm.texi: Regenerate.
31060 * targhooks.h (default_empty_mask_is_expensive): Declare.
31061 * targhooks.c (default_empty_mask_is_expensive): New function.
31062 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
31063 if the target says that empty masks are expensive.
31064 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
31066 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
31068 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31069 Alan Hayward <alan.hayward@arm.com>
31070 David Sherwood <david.sherwood@arm.com>
31072 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
31073 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
31074 (vect_use_loop_mask_for_alignment_p): New function.
31075 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
31076 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
31077 niters_skip argument. Make sure that the first niters_skip elements
31078 of the first iteration are inactive.
31079 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
31080 Update call to vect_set_loop_masks_directly.
31081 (get_misalign_in_elems): New function, split out from...
31082 (vect_gen_prolog_loop_niters): ...here.
31083 (vect_update_init_of_dr): Take a code argument that specifies whether
31084 the adjustment should be added or subtracted.
31085 (vect_update_init_of_drs): Likewise.
31086 (vect_prepare_for_masked_peels): New function.
31087 (vect_do_peeling): Skip prologue peeling if we're using a mask
31088 instead. Update call to vect_update_inits_of_drs.
31089 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
31091 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
31092 alignment. Do not include the number of peeled iterations in
31093 the minimum threshold in that case.
31094 (vectorizable_induction): Adjust the start value down by
31095 LOOP_VINFO_MASK_SKIP_NITERS iterations.
31096 (vect_transform_loop): Call vect_prepare_for_masked_peels.
31097 Take the number of skipped iterations into account when calculating
31099 * tree-vect-stmts.c (vect_gen_while_not): New function.
31101 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31102 Alan Hayward <alan.hayward@arm.com>
31103 David Sherwood <david.sherwood@arm.com>
31105 * doc/sourcebuild.texi (vect_fully_masked): Document.
31106 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
31107 default value to 0.
31108 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
31110 (vect_analyze_loop_2): ...here. Don't check the vectorization
31111 factor against the number of loop iterations if the loop is
31114 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31115 Alan Hayward <alan.hayward@arm.com>
31116 David Sherwood <david.sherwood@arm.com>
31118 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
31119 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
31120 (dump_groups): Update accordingly.
31121 (iv_use::mem_type): New member variable.
31122 (address_p): New function.
31123 (record_use): Add a mem_type argument and initialize the new
31125 (record_group_use): Add a mem_type argument. Use address_p.
31126 Remove obsolete null checks of base_object. Update call to record_use.
31127 (find_interesting_uses_op): Update call to record_group_use.
31128 (find_interesting_uses_cond): Likewise.
31129 (find_interesting_uses_address): Likewise.
31130 (get_mem_type_for_internal_fn): New function.
31131 (find_address_like_use): Likewise.
31132 (find_interesting_uses_stmt): Try find_address_like_use before
31133 calling find_interesting_uses_op.
31134 (addr_offset_valid_p): Use the iv mem_type field as the type
31135 of the addressed memory.
31136 (add_autoinc_candidates): Likewise.
31137 (get_address_cost): Likewise.
31138 (split_small_address_groups_p): Use address_p.
31139 (split_address_groups): Likewise.
31140 (add_iv_candidate_for_use): Likewise.
31141 (autoinc_possible_for_pair): Likewise.
31142 (rewrite_groups): Likewise.
31143 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
31144 (determine_group_iv_cost): Update after split of USE_ADDRESS.
31145 (get_alias_ptr_type_for_ptr_address): New function.
31146 (rewrite_use_address): Rewrite address uses in calls that were
31147 identified by find_address_like_use.
31149 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31150 Alan Hayward <alan.hayward@arm.com>
31151 David Sherwood <david.sherwood@arm.com>
31153 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
31155 * gimple-expr.h (is_gimple_addressable: Likewise.
31156 * gimple-expr.c (is_gimple_address): Likewise.
31157 * internal-fn.c (expand_call_mem_ref): New function.
31158 (expand_mask_load_optab_fn): Use it.
31159 (expand_mask_store_optab_fn): Likewise.
31161 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31162 Alan Hayward <alan.hayward@arm.com>
31163 David Sherwood <david.sherwood@arm.com>
31165 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
31166 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
31167 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
31168 (cond_umax@var{mode}): Document.
31169 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
31170 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
31171 (cond_umin_optab, cond_umax_optab): New optabs.
31172 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
31173 (COND_IOR, COND_XOR): New internal functions.
31174 * internal-fn.h (get_conditional_internal_fn): Declare.
31175 * internal-fn.c (cond_binary_direct): New macro.
31176 (expand_cond_binary_optab_fn): Likewise.
31177 (direct_cond_binary_optab_supported_p): Likewise.
31178 (get_conditional_internal_fn): New function.
31179 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
31180 Cope with reduction statements that are vectorized as calls rather
31182 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
31183 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
31184 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
31185 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
31186 (UNSPEC_COND_EOR): New unspecs.
31187 (optab): Add mappings for them.
31188 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
31189 (sve_int_op, sve_fp_op): New int attributes.
31191 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31192 Alan Hayward <alan.hayward@arm.com>
31193 David Sherwood <david.sherwood@arm.com>
31195 * optabs.def (while_ult_optab): New optab.
31196 * doc/md.texi (while_ult@var{m}@var{n}): Document.
31197 * internal-fn.def (WHILE_ULT): New internal function.
31198 * internal-fn.h (direct_internal_fn_supported_p): New override
31199 that takes two types as argument.
31200 * internal-fn.c (while_direct): New macro.
31201 (expand_while_optab_fn): New function.
31202 (convert_optab_supported_p): Likewise.
31203 (direct_while_optab_supported_p): New macro.
31204 * wide-int.h (wi::udiv_ceil): New function.
31205 * tree-vectorizer.h (rgroup_masks): New structure.
31206 (vec_loop_masks): New typedef.
31207 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
31208 and fully_masked_p.
31209 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
31210 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
31211 (vect_max_vf): New function.
31212 (slpeel_make_loop_iterate_ntimes): Delete.
31213 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
31214 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
31215 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
31216 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
31217 internal-fn.h, stor-layout.h and optabs-query.h.
31218 (vect_set_loop_mask): New function.
31219 (add_preheader_seq): Likewise.
31220 (add_header_seq): Likewise.
31221 (interleave_supported_p): Likewise.
31222 (vect_maybe_permute_loop_masks): Likewise.
31223 (vect_set_loop_masks_directly): Likewise.
31224 (vect_set_loop_condition_masked): Likewise.
31225 (vect_set_loop_condition_unmasked): New function, split out from
31226 slpeel_make_loop_iterate_ntimes.
31227 (slpeel_make_loop_iterate_ntimes): Rename to..
31228 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
31229 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
31230 (vect_do_peeling): Update call accordingly.
31231 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
31233 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
31234 mask_compare_type, can_fully_mask_p and fully_masked_p.
31235 (release_vec_loop_masks): New function.
31236 (_loop_vec_info): Use it to free the loop masks.
31237 (can_produce_all_loop_masks_p): New function.
31238 (vect_get_max_nscalars_per_iter): Likewise.
31239 (vect_verify_full_masking): Likewise.
31240 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
31241 retries, and free the mask rgroups before retrying. Check loop-wide
31242 reasons for disallowing fully-masked loops. Make the final decision
31243 about whether use a fully-masked loop or not.
31244 (vect_estimate_min_profitable_iters): Do not assume that peeling
31245 for the number of iterations will be needed for fully-masked loops.
31246 (vectorizable_reduction): Disable fully-masked loops.
31247 (vectorizable_live_operation): Likewise.
31248 (vect_halve_mask_nunits): New function.
31249 (vect_double_mask_nunits): Likewise.
31250 (vect_record_loop_mask): Likewise.
31251 (vect_get_loop_mask): Likewise.
31252 (vect_transform_loop): Handle the case in which the final loop
31253 iteration might handle a partial vector. Call vect_set_loop_condition
31254 instead of slpeel_make_loop_iterate_ntimes.
31255 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
31256 (check_load_store_masking): New function.
31257 (prepare_load_store_mask): Likewise.
31258 (vectorizable_store): Handle fully-masked loops.
31259 (vectorizable_load): Likewise.
31260 (supportable_widening_operation): Use vect_halve_mask_nunits for
31262 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
31263 (vect_gen_while): New function.
31264 * config/aarch64/aarch64.md (umax<mode>3): New expander.
31265 (aarch64_uqdec<mode>): New insn.
31267 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31268 Alan Hayward <alan.hayward@arm.com>
31269 David Sherwood <david.sherwood@arm.com>
31271 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
31272 (reduc_xor_scal_optab): New optabs.
31273 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
31274 (reduc_xor_scal_@var{m}): Document.
31275 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
31276 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
31277 internal functions.
31278 * fold-const-call.c (fold_const_call): Handle them.
31279 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
31280 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
31281 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
31282 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
31283 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
31284 (UNSPEC_XORV): New unspecs.
31285 (optab): Add entries for them.
31286 (BITWISEV): New int iterator.
31287 (bit_reduc_op): New int attributes.
31289 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31290 Alan Hayward <alan.hayward@arm.com>
31291 David Sherwood <david.sherwood@arm.com>
31293 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
31294 * internal-fn.def (VEC_SHL_INSERT): New internal function.
31295 * optabs.def (vec_shl_insert_optab): New optab.
31296 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
31297 (duplicate_and_interleave): Likewise.
31298 * tree-vect-loop.c: Include internal-fn.h.
31299 (neutral_op_for_slp_reduction): New function, split out from
31300 get_initial_defs_for_reduction.
31301 (get_initial_def_for_reduction): Handle option 2 for variable-length
31302 vectors by loading the neutral value into a vector and then shifting
31303 the initial value into element 0.
31304 (get_initial_defs_for_reduction): Replace the code argument with
31305 the neutral value calculated by neutral_op_for_slp_reduction.
31306 Use gimple_build_vector for constant-length vectors.
31307 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
31308 but the first group_size elements have a neutral value.
31309 Use duplicate_and_interleave otherwise.
31310 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
31311 Update call to get_initial_defs_for_reduction. Handle SLP
31312 reductions for variable-length vectors by creating one vector
31313 result for each scalar result, with the elements associated
31314 with other scalar results stubbed out with the neutral value.
31315 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
31316 Require IFN_VEC_SHL_INSERT for double reductions on
31317 variable-length vectors, or SLP reductions that have
31318 a neutral value. Require can_duplicate_and_interleave_p
31319 support for variable-length unchained SLP reductions if there
31320 is no neutral value, such as for MIN/MAX reductions. Also require
31321 the number of vector elements to be a multiple of the number of
31322 SLP statements when doing variable-length unchained SLP reductions.
31323 Update call to vect_create_epilog_for_reduction.
31324 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
31325 and remove initial values.
31326 (duplicate_and_interleave): Make public.
31327 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
31328 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
31330 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31331 Alan Hayward <alan.hayward@arm.com>
31332 David Sherwood <david.sherwood@arm.com>
31334 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
31335 (can_duplicate_and_interleave_p): New function.
31336 (vect_get_and_check_slp_defs): Take the vector of statements
31337 rather than just the current one. Remove excess parentheses.
31338 Restriction rejectinon of vect_constant_def and vect_external_def
31339 for variable-length vectors to boolean types, or types for which
31340 can_duplicate_and_interleave_p is false.
31341 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
31342 (duplicate_and_interleave): New function.
31343 (vect_get_constant_vectors): Use gimple_build_vector for
31344 constant-length vectors and suitable variable-length constant
31345 vectors. Use duplicate_and_interleave for other variable-length
31346 vectors. Don't defer the update when inserting new statements.
31348 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31349 Alan Hayward <alan.hayward@arm.com>
31350 David Sherwood <david.sherwood@arm.com>
31352 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
31353 min_profitable_iters doesn't go negative.
31355 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31356 Alan Hayward <alan.hayward@arm.com>
31357 David Sherwood <david.sherwood@arm.com>
31359 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
31360 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
31361 * optabs.def (vec_mask_load_lanes_optab): New optab.
31362 (vec_mask_store_lanes_optab): Likewise.
31363 * internal-fn.def (MASK_LOAD_LANES): New internal function.
31364 (MASK_STORE_LANES): Likewise.
31365 * internal-fn.c (mask_load_lanes_direct): New macro.
31366 (mask_store_lanes_direct): Likewise.
31367 (expand_mask_load_optab_fn): Handle masked operations.
31368 (expand_mask_load_lanes_optab_fn): New macro.
31369 (expand_mask_store_optab_fn): Handle masked operations.
31370 (expand_mask_store_lanes_optab_fn): New macro.
31371 (direct_mask_load_lanes_optab_supported_p): Likewise.
31372 (direct_mask_store_lanes_optab_supported_p): Likewise.
31373 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
31375 (vect_load_lanes_supported): Likewise.
31376 * tree-vect-data-refs.c (strip_conversion): New function.
31377 (can_group_stmts_p): Likewise.
31378 (vect_analyze_data_ref_accesses): Use it instead of checking
31379 for a pair of assignments.
31380 (vect_store_lanes_supported): Take a masked_p parameter.
31381 (vect_load_lanes_supported): Likewise.
31382 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
31383 vect_store_lanes_supported and vect_load_lanes_supported.
31384 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
31385 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
31386 parameter. Don't allow gaps for masked accesses.
31387 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
31388 and vect_load_lanes_supported.
31389 (get_load_store_type): Take a masked_p parameter and update
31390 call to get_group_load_store_type.
31391 (vectorizable_store): Update call to get_load_store_type.
31392 Handle IFN_MASK_STORE_LANES.
31393 (vectorizable_load): Update call to get_load_store_type.
31394 Handle IFN_MASK_LOAD_LANES.
31396 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31397 Alan Hayward <alan.hayward@arm.com>
31398 David Sherwood <david.sherwood@arm.com>
31400 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
31402 * config/aarch64/aarch64-protos.h
31403 (aarch64_sve_struct_memory_operand_p): Declare.
31404 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
31405 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
31406 (VPRED, vpred): Handle SVE structure modes.
31407 * config/aarch64/constraints.md (Utx): New constraint.
31408 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
31409 (aarch64_sve_struct_nonimmediate_operand): New predicates.
31410 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
31411 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
31412 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
31413 structure modes. Split into pieces after RA.
31414 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
31415 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
31417 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
31418 SVE structure modes.
31419 (aarch64_classify_address): Likewise.
31420 (sizetochar): Move earlier in file.
31421 (aarch64_print_operand): Handle SVE register lists.
31422 (aarch64_array_mode): New function.
31423 (aarch64_sve_struct_memory_operand_p): Likewise.
31424 (TARGET_ARRAY_MODE): Redefine.
31426 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31427 Alan Hayward <alan.hayward@arm.com>
31428 David Sherwood <david.sherwood@arm.com>
31430 * target.def (array_mode): New target hook.
31431 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
31432 * doc/tm.texi: Regenerate.
31433 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
31434 * hooks.c (hook_optmode_mode_uhwi_none): New function.
31435 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
31436 targetm.array_mode.
31437 * stor-layout.c (mode_for_array): Likewise. Support polynomial
31440 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31441 Alan Hayward <alan.hayward@arm.com>
31442 David Sherwood <david.sherwood@arm.com>
31444 * fold-const.c (fold_binary_loc): Check the argument types
31445 rather than the result type when testing for a vector operation.
31447 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31449 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
31450 * doc/tm.texi: Regenerate.
31452 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31453 Alan Hayward <alan.hayward@arm.com>
31454 David Sherwood <david.sherwood@arm.com>
31456 * doc/invoke.texi (-msve-vector-bits=): Document new option.
31457 (sve): Document new AArch64 extension.
31458 * doc/md.texi (w): Extend the description of the AArch64
31459 constraint to include SVE vectors.
31460 (Upl, Upa): Document new AArch64 predicate constraints.
31461 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
31463 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
31464 (msve-vector-bits=): New option.
31465 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
31466 SVE when these are disabled.
31467 (sve): New extension.
31468 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
31469 modes. Adjust their number of units based on aarch64_sve_vg.
31470 (MAX_BITSIZE_MODE_ANY_MODE): Define.
31471 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
31472 aarch64_addr_query_type.
31473 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
31474 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
31475 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
31476 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
31477 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
31478 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
31479 (aarch64_simd_imm_zero_p): Delete.
31480 (aarch64_check_zero_based_sve_index_immediate): Declare.
31481 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
31482 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
31483 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
31484 (aarch64_sve_float_mul_immediate_p): Likewise.
31485 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
31486 rather than an rtx.
31487 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
31488 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
31489 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
31490 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
31491 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
31492 (aarch64_regmode_natural_size): Likewise.
31493 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
31494 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
31496 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
31497 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
31498 for VG and the SVE predicate registers.
31499 (V_ALIASES): Add a "z"-prefixed alias.
31500 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
31501 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
31502 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
31503 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
31504 (REG_CLASS_NAMES): Add entries for them.
31505 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
31506 and the predicate registers.
31507 (aarch64_sve_vg): Declare.
31508 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
31509 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
31510 (REGMODE_NATURAL_SIZE): Define.
31511 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
31513 * config/aarch64/aarch64.c: Include cfgrtl.h.
31514 (simd_immediate_info): Add a constructor for series vectors,
31515 and an associated step field.
31516 (aarch64_sve_vg): New variable.
31517 (aarch64_dbx_register_number): Handle VG and the predicate registers.
31518 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
31519 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
31520 (VEC_ANY_DATA, VEC_STRUCT): New constants.
31521 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
31522 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
31523 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
31524 (aarch64_get_mask_mode): New functions.
31525 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
31526 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
31527 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
31528 predicate modes and predicate registers. Explicitly restrict
31529 GPRs to modes of 16 bytes or smaller. Only allow FP registers
31530 to store a vector mode if it is recognized by
31531 aarch64_classify_vector_mode.
31532 (aarch64_regmode_natural_size): New function.
31533 (aarch64_hard_regno_caller_save_mode): Return the original mode
31535 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
31536 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
31537 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
31538 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
31540 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
31541 does not overlap dest if the function is frame-related. Handle
31543 (aarch64_split_add_offset): New function.
31544 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
31545 them aarch64_add_offset.
31546 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
31547 and update call to aarch64_sub_sp.
31548 (aarch64_add_cfa_expression): New function.
31549 (aarch64_expand_prologue): Pass extra temporary registers to the
31550 functions above. Handle the case in which we need to emit new
31551 DW_CFA_expressions for registers that were originally saved
31552 relative to the stack pointer, but now have to be expressed
31553 relative to the frame pointer.
31554 (aarch64_output_mi_thunk): Pass extra temporary registers to the
31556 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
31557 IP0 and IP1 values for SVE frames.
31558 (aarch64_expand_vec_series): New function.
31559 (aarch64_expand_sve_widened_duplicate): Likewise.
31560 (aarch64_expand_sve_const_vector): Likewise.
31561 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
31562 Handle SVE constants. Use emit_move_insn to move a force_const_mem
31563 into the register, rather than emitting a SET directly.
31564 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
31565 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
31566 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
31567 (offset_9bit_signed_scaled_p): New functions.
31568 (aarch64_replicate_bitmask_imm): New function.
31569 (aarch64_bitmask_imm): Use it.
31570 (aarch64_cannot_force_const_mem): Reject expressions involving
31571 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
31572 (aarch64_classify_index): Handle SVE indices, by requiring
31573 a plain register index with a scale that matches the element size.
31574 (aarch64_classify_address): Handle SVE addresses. Assert that
31575 the mode of the address is VOIDmode or an integer mode.
31576 Update call to aarch64_classify_symbol.
31577 (aarch64_classify_symbolic_expression): Update call to
31578 aarch64_classify_symbol.
31579 (aarch64_const_vec_all_in_range_p): New function.
31580 (aarch64_print_vector_float_operand): Likewise.
31581 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
31582 "vN" for FP registers with SVE modes. Handle (const ...) vectors
31583 and the FP immediates 1.0 and 0.5.
31584 (aarch64_print_address_internal): Handle SVE addresses.
31585 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
31586 (aarch64_regno_regclass): Handle predicate registers.
31587 (aarch64_secondary_reload): Handle big-endian reloads of SVE
31589 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
31590 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
31591 (aarch64_convert_sve_vector_bits): New function.
31592 (aarch64_override_options): Use it to handle -msve-vector-bits=.
31593 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
31594 rather than an rtx.
31595 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
31596 Handle SVE vector and predicate modes. Accept VL-based constants
31597 that need only one temporary register, and VL offsets that require
31598 no temporary registers.
31599 (aarch64_conditional_register_usage): Mark the predicate registers
31600 as fixed if SVE isn't available.
31601 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
31602 Return true for SVE vector and predicate modes.
31603 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
31604 rather than an unsigned int. Handle SVE modes.
31605 (aarch64_preferred_simd_mode): Update call accordingly. Handle
31607 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
31609 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
31610 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
31611 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
31612 (aarch64_sve_float_mul_immediate_p): New functions.
31613 (aarch64_sve_valid_immediate): New function.
31614 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
31615 Explicitly reject structure modes. Check for INDEX constants.
31616 Handle PTRUE and PFALSE constants.
31617 (aarch64_check_zero_based_sve_index_immediate): New function.
31618 (aarch64_simd_imm_zero_p): Delete.
31619 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
31620 vector modes. Accept constants in the range of CNT[BHWD].
31621 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
31622 ask for an Advanced SIMD mode.
31623 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
31624 (aarch64_simd_vector_alignment): Handle SVE predicates.
31625 (aarch64_vectorize_preferred_vector_alignment): New function.
31626 (aarch64_simd_vector_alignment_reachable): Use it instead of
31628 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
31629 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
31631 (MAX_VECT_LEN): Delete.
31632 (expand_vec_perm_d): Add a vec_flags field.
31633 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
31634 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
31635 (aarch64_evpc_ext): Don't apply a big-endian lane correction
31637 (aarch64_evpc_rev): Rename to...
31638 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
31639 (aarch64_evpc_rev_global): New function.
31640 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
31641 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
31643 (aarch64_evpc_sve_tbl): New function.
31644 (aarch64_expand_vec_perm_const_1): Update after rename of
31645 aarch64_evpc_rev. Handle SVE permutes too, trying
31646 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
31647 than aarch64_evpc_tbl.
31648 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
31649 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
31650 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
31651 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
31652 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
31653 (aarch64_expand_sve_vcond): New functions.
31654 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
31655 of aarch64_vector_mode_p.
31656 (aarch64_dwarf_poly_indeterminate_value): New function.
31657 (aarch64_compute_pressure_classes): Likewise.
31658 (aarch64_can_change_mode_class): Likewise.
31659 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
31660 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
31661 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
31662 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
31663 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
31664 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
31665 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
31666 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
31668 (Dn, Dl, Dr): Accept const as well as const_vector.
31669 (Dz): Likewise. Compare against CONST0_RTX.
31670 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
31671 of "vector" where appropriate.
31672 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
31673 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
31674 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
31675 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
31676 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
31677 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
31678 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
31679 (v_int_equiv): Extend to SVE modes.
31680 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
31682 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
31683 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
31684 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
31685 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
31686 (SVE_COND_FP_CMP): New int iterators.
31687 (perm_hilo): Handle the new unpack unspecs.
31688 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
31690 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
31691 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
31692 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
31693 (aarch64_equality_operator, aarch64_constant_vector_operand)
31694 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
31695 (aarch64_sve_nonimmediate_operand): Likewise.
31696 (aarch64_sve_general_operand): Likewise.
31697 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
31698 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
31699 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
31700 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
31701 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
31702 (aarch64_sve_float_arith_immediate): Likewise.
31703 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
31704 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
31705 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
31706 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
31707 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
31708 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
31709 (aarch64_sve_float_arith_operand): Likewise.
31710 (aarch64_sve_float_arith_with_sub_operand): Likewise.
31711 (aarch64_sve_float_mul_operand): Likewise.
31712 (aarch64_sve_vec_perm_operand): Likewise.
31713 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
31714 (aarch64_mov_operand): Accept const_poly_int and const_vector.
31715 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
31716 as well as const_vector.
31717 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
31718 in file. Use CONST0_RTX and CONSTM1_RTX.
31719 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
31720 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
31721 Use aarch64_simd_imm_zero.
31722 * config/aarch64/aarch64-sve.md: New file.
31723 * config/aarch64/aarch64.md: Include it.
31724 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
31725 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
31726 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
31727 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
31728 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
31729 (sve): New attribute.
31730 (enabled): Disable instructions with the sve attribute unless
31732 (movqi, movhi): Pass CONST_POLY_INT operaneds through
31733 aarch64_expand_mov_immediate.
31734 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
31735 CNT[BHSD] immediates.
31736 (movti): Split CONST_POLY_INT moves into two halves.
31737 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
31738 Split additions that need a temporary here if the destination
31739 is the stack pointer.
31740 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
31741 (*add<mode>3_poly_1): New instruction.
31742 (set_clobber_cc): New expander.
31744 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31746 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
31747 parameter and use it instead of GET_MODE_SIZE (innermode). Use
31748 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
31749 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
31750 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
31751 Change innermode from fixed_mode_size to machine_mode.
31752 (simplify_subreg): Update call accordingly. Handle a constant-sized
31753 subreg of a variable-length CONST_VECTOR.
31755 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31756 Alan Hayward <alan.hayward@arm.com>
31757 David Sherwood <david.sherwood@arm.com>
31759 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
31760 (add_offset_to_base): New function, split out from...
31761 (create_mem_ref): ...here. When handling a scale other than 1,
31762 check first whether the address is valid without the offset.
31763 Add it into the base if so, leaving the index and scale as-is.
31765 2018-01-12 Jakub Jelinek <jakub@redhat.com>
31768 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
31769 fold_for_warn before checking if arg2 is INTEGER_CST.
31771 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
31773 * config/rs6000/predicates.md (load_multiple_operation): Delete.
31774 (store_multiple_operation): Delete.
31775 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
31776 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
31777 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
31778 guarded by TARGET_STRING.
31779 (rs6000_output_load_multiple): Delete.
31780 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
31781 OPTION_MASK_STRING / TARGET_STRING handling.
31782 (print_operand) <'N', 'O'>: Add comment that these are unused now.
31783 (const rs6000_opt_masks) <"string">: Change mask to 0.
31784 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
31785 (MASK_STRING): Delete.
31786 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
31788 (load_multiple): Delete.
31795 (store_multiple): Delete.
31802 (movmemsi_8reg): Delete.
31803 (corresponding unnamed define_insn): Delete.
31804 (movmemsi_6reg): Delete.
31805 (corresponding unnamed define_insn): Delete.
31806 (movmemsi_4reg): Delete.
31807 (corresponding unnamed define_insn): Delete.
31808 (movmemsi_2reg): Delete.
31809 (corresponding unnamed define_insn): Delete.
31810 (movmemsi_1reg): Delete.
31811 (corresponding unnamed define_insn): Delete.
31812 * config/rs6000/rs6000.opt (mno-string): New.
31813 (mstring): Replace by deprecation warning stub.
31814 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
31816 2018-01-12 Jakub Jelinek <jakub@redhat.com>
31818 * regrename.c (regrename_do_replace): If replacing the same
31819 reg multiple times, try to reuse last created gen_raw_REG.
31822 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
31823 main to workaround a bug in GDB.
31825 2018-01-12 Tom de Vries <tom@codesourcery.com>
31828 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
31830 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
31832 PR rtl-optimization/80481
31833 * ira-color.c (get_cap_member): New function.
31834 (allocnos_conflict_by_live_ranges_p): Use it.
31835 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
31836 (setup_slot_coalesced_allocno_live_ranges): Ditto.
31838 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
31841 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
31842 (*saddl_se_1): Ditto.
31843 (*ssubsi_1): Ditto.
31844 (*ssubl_se_1): Ditto.
31846 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
31848 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
31849 rather than wi::to_widest for DR_INITs.
31850 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
31851 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
31852 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
31854 (vect_analyze_group_access_1): Note that here.
31856 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
31858 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
31859 polynomial type sizes.
31861 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
31863 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
31864 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
31865 (gimple_add_tmp_var): Likewise.
31867 2018-01-12 Martin Liska <mliska@suse.cz>
31869 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
31870 (gimple_alloc_sizes): Likewise.
31871 (dump_gimple_statistics): Use PRIu64 in printf format.
31872 * gimple.h: Change uint64_t to int.
31874 2018-01-12 Martin Liska <mliska@suse.cz>
31876 * tree-core.h: Use uint64_t instead of int.
31877 * tree.c (tree_node_counts): Likewise.
31878 (tree_node_sizes): Likewise.
31879 (dump_tree_statistics): Use PRIu64 in printf format.
31881 2018-01-12 Martin Liska <mliska@suse.cz>
31883 * Makefile.in: As qsort_chk is implemented in vec.c, add
31884 vec.o to linkage of gencfn-macros.
31885 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
31886 passing the info to record_node_allocation_statistics.
31887 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
31889 * ggc-common.c (struct ggc_usage): Add operator== and use
31890 it in operator< and compare function.
31891 * mem-stats.h (struct mem_usage): Likewise.
31892 * vec.c (struct vec_usage): Remove operator< and compare
31893 function. Can be simply inherited.
31895 2018-01-12 Martin Jambor <mjambor@suse.cz>
31898 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
31899 * tree-ssa-math-opts.c: Include domwalk.h.
31900 (convert_mult_to_fma_1): New function.
31901 (fma_transformation_info): New type.
31902 (fma_deferring_state): Likewise.
31903 (cancel_fma_deferring): New function.
31904 (result_of_phi): Likewise.
31905 (last_fma_candidate_feeds_initial_phi): Likewise.
31906 (convert_mult_to_fma): Added deferring logic, split actual
31907 transformation to convert_mult_to_fma_1.
31908 (math_opts_dom_walker): New type.
31909 (math_opts_dom_walker::after_dom_children): New method, body moved
31910 here from pass_optimize_widening_mul::execute, added deferring logic
31912 (pass_optimize_widening_mul::execute): Moved most of code to
31913 math_opts_dom_walker::after_dom_children.
31914 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
31915 * config/i386/i386.c (ix86_option_override_internal): Added
31916 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
31918 2018-01-12 Richard Biener <rguenther@suse.de>
31921 * dwarf2out.c (gen_variable_die): Do not reset old_die for
31922 inline instance vars.
31924 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
31927 * config/rx/rx.c (rx_is_restricted_memory_address):
31928 Handle SUBREG case.
31930 2018-01-12 Richard Biener <rguenther@suse.de>
31932 PR tree-optimization/80846
31933 * target.def (split_reduction): New target hook.
31934 * targhooks.c (default_split_reduction): New function.
31935 * targhooks.h (default_split_reduction): Declare.
31936 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
31937 target requests first reduce vectors by combining low and high
31939 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
31940 (get_vectype_for_scalar_type_and_size): Export.
31941 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
31942 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
31943 * doc/tm.texi: Regenerate.
31944 * config/i386/i386.c (ix86_split_reduction): Implement
31945 TARGET_VECTORIZE_SPLIT_REDUCTION.
31947 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
31950 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
31951 in PIC mode except for TARGET_VXWORKS_RTP.
31952 * config/sparc/sparc.c: Include cfgrtl.h.
31953 (TARGET_INIT_PIC_REG): Define.
31954 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
31955 (sparc_pic_register_p): New predicate.
31956 (sparc_legitimate_address_p): Use it.
31957 (sparc_legitimize_pic_address): Likewise.
31958 (sparc_delegitimize_address): Likewise.
31959 (sparc_mode_dependent_address_p): Likewise.
31960 (gen_load_pcrel_sym): Remove 4th parameter.
31961 (load_got_register): Adjust call to above. Remove obsolete stuff.
31962 (sparc_expand_prologue): Do not call load_got_register here.
31963 (sparc_flat_expand_prologue): Likewise.
31964 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
31965 (sparc_use_pseudo_pic_reg): New function.
31966 (sparc_init_pic_reg): Likewise.
31967 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
31968 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
31970 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
31972 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
31973 Add item for branch_cost.
31975 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
31977 PR rtl-optimization/83565
31978 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
31979 not extend the result to a larger mode for rotate operations.
31980 (num_sign_bit_copies1): Likewise.
31982 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31985 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
31987 Use values-Xc.o for -pedantic.
31988 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
31990 2018-01-12 Martin Liska <mliska@suse.cz>
31993 * ipa-devirt.c (final_warning_record::grow_type_warnings):
31995 (possible_polymorphic_call_targets): Use it.
31996 (ipa_devirt): Likewise.
31998 2018-01-12 Martin Liska <mliska@suse.cz>
32000 * profile-count.h (enum profile_quality): Use 0 as invalid
32001 enum value of profile_quality.
32003 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
32005 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
32006 -mext-string options.
32008 2018-01-12 Richard Biener <rguenther@suse.de>
32010 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
32011 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
32012 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
32014 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
32016 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
32018 * configure.ac (--with-long-double-format): Add support for the
32019 configuration option to change the default long double format on
32021 * config.gcc (powerpc*-linux*-*): Likewise.
32022 * configure: Regenerate.
32023 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
32024 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
32025 used without modification.
32027 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32029 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
32030 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
32031 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
32032 MISC_BUILTIN_SPEC_BARRIER.
32033 (rs6000_init_builtins): Likewise.
32034 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
32036 (speculation_barrier): New define_insn.
32037 * doc/extend.texi: Document __builtin_speculation_barrier.
32039 2018-01-11 Jakub Jelinek <jakub@redhat.com>
32042 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
32043 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
32044 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
32046 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
32047 integral modes instead of "ss" and "sd".
32048 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
32049 vectors with 32-bit and 64-bit elements.
32050 (vecdupssescalarmodesuffix): New mode attribute.
32051 (vec_dup<mode>): Use it.
32053 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
32056 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
32057 frame if argument is passed on stack.
32059 2018-01-11 Jakub Jelinek <jakub@redhat.com>
32062 * ree.c (combine_reaching_defs): Optimize also
32063 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
32064 reg2=any_extend(exp); reg1=reg2;, formatting fix.
32066 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
32068 PR middle-end/83189
32069 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
32071 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
32073 PR middle-end/83718
32074 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
32075 after they are computed.
32077 2018-01-11 Bin Cheng <bin.cheng@arm.com>
32079 PR tree-optimization/83695
32080 * gimple-loop-linterchange.cc
32081 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
32082 reset cached scev information after interchange.
32083 (pass_linterchange::execute): Remove call to scev_reset_htab.
32085 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32087 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
32088 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
32089 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
32090 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
32091 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
32092 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
32093 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
32094 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
32095 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
32096 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
32097 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
32098 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
32099 (V_lane_reg): Likewise.
32100 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
32102 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
32103 (vfmal_lane_low<mode>_intrinsic,
32104 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
32105 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
32106 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
32107 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
32108 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
32109 vfmsl_lane_high<mode>_intrinsic): New define_insns.
32111 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32113 * config/arm/arm-cpus.in (fp16fml): New feature.
32114 (ALL_SIMD): Add fp16fml.
32115 (armv8.2-a): Add fp16fml as an option.
32116 (armv8.3-a): Likewise.
32117 (armv8.4-a): Add fp16fml as part of fp16.
32118 * config/arm/arm.h (TARGET_FP16FML): Define.
32119 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
32121 * config/arm/arm-modes.def (V2HF): Define.
32122 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
32123 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
32124 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
32125 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
32126 vfmsl_low, vfmsl_high): New set of builtins.
32127 * config/arm/iterators.md (PLUSMINUS): New code iterator.
32128 (vfml_op): New code attribute.
32129 (VFMLHALVES): New int iterator.
32130 (VFML, VFMLSEL): New mode attributes.
32131 (V_reg): Define mapping for V2HF.
32132 (V_hi, V_lo): New mode attributes.
32133 (VF_constraint): Likewise.
32134 (vfml_half, vfml_half_selector): New int attributes.
32135 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
32137 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
32138 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
32140 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
32141 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
32142 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
32143 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
32145 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
32146 Document new effective target and option set.
32148 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32150 * config/arm/arm-cpus.in (armv8_4): New feature.
32151 (ARMv8_4a): New fgroup.
32152 (armv8.4-a): New arch.
32153 * config/arm/arm-tables.opt: Regenerate.
32154 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
32155 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
32156 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
32157 Add matching rules for -march=armv8.4-a and extensions.
32158 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
32160 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
32163 * config/rx/rx.md (BW): New mode attribute.
32164 (sync_lock_test_and_setsi): Add mode suffix to insn output.
32166 2018-01-11 Richard Biener <rguenther@suse.de>
32168 PR tree-optimization/83435
32169 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
32170 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
32171 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
32173 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32174 Alan Hayward <alan.hayward@arm.com>
32175 David Sherwood <david.sherwood@arm.com>
32177 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
32179 (aarch64_classify_address): Initialize it. Track polynomial offsets.
32180 (aarch64_print_address_internal): Use it to check for a zero offset.
32182 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32183 Alan Hayward <alan.hayward@arm.com>
32184 David Sherwood <david.sherwood@arm.com>
32186 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
32187 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
32188 Return a poly_int64 rather than a HOST_WIDE_INT.
32189 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
32190 rather than a HOST_WIDE_INT.
32191 * config/aarch64/aarch64.h (aarch64_frame): Protect with
32192 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
32193 hard_fp_offset, frame_size, initial_adjust, callee_offset and
32194 final_offset from HOST_WIDE_INT to poly_int64.
32195 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
32196 to_constant when getting the number of units in an Advanced SIMD
32198 (aarch64_builtin_vectorized_function): Check for a constant number
32200 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
32202 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
32203 attribute instead of GET_MODE_NUNITS.
32204 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
32205 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
32206 GET_MODE_SIZE for fixed-size registers.
32207 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
32208 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
32209 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
32210 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
32211 (aarch64_print_operand, aarch64_print_address_internal)
32212 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
32213 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
32214 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
32215 Handle polynomial GET_MODE_SIZE.
32216 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
32217 wider than SImode without modification.
32218 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
32219 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
32220 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
32221 passing and returning SVE modes.
32222 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
32223 rather than GEN_INT.
32224 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
32225 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
32226 (aarch64_allocate_and_probe_stack_space): Likewise.
32227 (aarch64_layout_frame): Cope with polynomial offsets.
32228 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
32229 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
32230 polynomial offsets.
32231 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
32232 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
32233 poly_int64 rather than a HOST_WIDE_INT.
32234 (aarch64_get_separate_components, aarch64_process_components)
32235 (aarch64_expand_prologue, aarch64_expand_epilogue)
32236 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
32237 (aarch64_anchor_offset): New function, split out from...
32238 (aarch64_legitimize_address): ...here.
32239 (aarch64_builtin_vectorization_cost): Handle polynomial
32240 TYPE_VECTOR_SUBPARTS.
32241 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
32243 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
32244 number of elements from the PARALLEL rather than the mode.
32245 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
32246 rather than GET_MODE_BITSIZE.
32247 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
32248 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
32249 (aarch64_expand_vec_perm_const_1): Handle polynomial
32250 d->perm.length () and d->perm elements.
32251 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
32252 Apply to_constant to d->perm elements.
32253 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
32254 polynomial CONST_VECTOR_NUNITS.
32255 (aarch64_move_pointer): Take amount as a poly_int64 rather
32257 (aarch64_progress_pointer): Avoid temporary variable.
32258 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
32259 the mode attribute instead of GET_MODE.
32261 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32262 Alan Hayward <alan.hayward@arm.com>
32263 David Sherwood <david.sherwood@arm.com>
32265 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
32266 x exists before using it.
32267 (aarch64_add_constant_internal): Rename to...
32268 (aarch64_add_offset_1): ...this. Replace regnum with separate
32269 src and dest rtxes. Handle the case in which they're different,
32270 including when the offset is zero. Replace scratchreg with an rtx.
32271 Use 2 additions if there is no spare register into which we can
32272 move a 16-bit constant.
32273 (aarch64_add_constant): Delete.
32274 (aarch64_add_offset): Replace reg with separate src and dest
32275 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
32276 Use aarch64_add_offset_1.
32277 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
32278 an rtx rather than an int. Take the delta as a poly_int64
32279 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
32280 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
32281 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
32282 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
32283 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
32284 and aarch64_add_sp.
32285 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
32286 aarch64_add_constant.
32288 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32290 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
32291 Use scalar_float_mode.
32293 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32295 * config/aarch64/aarch64-simd.md
32296 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
32297 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
32298 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
32299 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
32300 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
32301 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
32302 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
32303 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
32304 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
32305 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
32307 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32310 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
32311 targ_options->x_arm_arch_string is non NULL.
32313 2018-01-11 Tamar Christina <tamar.christina@arm.com>
32315 * config/aarch64/aarch64.h
32316 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
32318 2018-01-11 Sudakshina Das <sudi.das@arm.com>
32321 * expmed.c (emit_store_flag_force): Swap if const op0
32322 and change VOIDmode to mode of op0.
32324 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32326 PR rtl-optimization/83761
32327 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
32328 than bytes to mode_for_size.
32330 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
32332 PR middle-end/83189
32333 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
32334 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
32337 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
32339 PR middle-end/83575
32340 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
32341 when in layout mode.
32342 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
32343 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
32346 2018-01-10 Michael Collison <michael.collison@arm.com>
32348 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
32349 * config/aarch64/aarch64-option-extension.def: Add
32350 AARCH64_OPT_EXTENSION of 'fp16fml'.
32351 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
32352 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
32353 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
32354 * config/aarch64/constraints.md (Ui7): New constraint.
32355 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
32356 (VFMLA_SEL_W): Ditto.
32359 (VFMLA16_LOW): New int iterator.
32360 (VFMLA16_HIGH): Ditto.
32361 (UNSPEC_FMLAL): New unspec.
32362 (UNSPEC_FMLSL): Ditto.
32363 (UNSPEC_FMLAL2): Ditto.
32364 (UNSPEC_FMLSL2): Ditto.
32365 (f16mac): New code attribute.
32366 * config/aarch64/aarch64-simd-builtins.def
32367 (aarch64_fmlal_lowv2sf): Ditto.
32368 (aarch64_fmlsl_lowv2sf): Ditto.
32369 (aarch64_fmlalq_lowv4sf): Ditto.
32370 (aarch64_fmlslq_lowv4sf): Ditto.
32371 (aarch64_fmlal_highv2sf): Ditto.
32372 (aarch64_fmlsl_highv2sf): Ditto.
32373 (aarch64_fmlalq_highv4sf): Ditto.
32374 (aarch64_fmlslq_highv4sf): Ditto.
32375 (aarch64_fmlal_lane_lowv2sf): Ditto.
32376 (aarch64_fmlsl_lane_lowv2sf): Ditto.
32377 (aarch64_fmlal_laneq_lowv2sf): Ditto.
32378 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
32379 (aarch64_fmlalq_lane_lowv4sf): Ditto.
32380 (aarch64_fmlsl_lane_lowv4sf): Ditto.
32381 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
32382 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
32383 (aarch64_fmlal_lane_highv2sf): Ditto.
32384 (aarch64_fmlsl_lane_highv2sf): Ditto.
32385 (aarch64_fmlal_laneq_highv2sf): Ditto.
32386 (aarch64_fmlsl_laneq_highv2sf): Ditto.
32387 (aarch64_fmlalq_lane_highv4sf): Ditto.
32388 (aarch64_fmlsl_lane_highv4sf): Ditto.
32389 (aarch64_fmlalq_laneq_highv4sf): Ditto.
32390 (aarch64_fmlsl_laneq_highv4sf): Ditto.
32391 * config/aarch64/aarch64-simd.md:
32392 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
32393 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
32394 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
32395 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
32396 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
32397 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
32398 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
32399 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
32400 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
32401 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
32402 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
32403 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
32404 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
32405 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
32406 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
32407 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
32408 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
32409 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
32410 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
32411 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
32412 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
32413 (vfmlsl_low_u32): Ditto.
32414 (vfmlalq_low_u32): Ditto.
32415 (vfmlslq_low_u32): Ditto.
32416 (vfmlal_high_u32): Ditto.
32417 (vfmlsl_high_u32): Ditto.
32418 (vfmlalq_high_u32): Ditto.
32419 (vfmlslq_high_u32): Ditto.
32420 (vfmlal_lane_low_u32): Ditto.
32421 (vfmlsl_lane_low_u32): Ditto.
32422 (vfmlal_laneq_low_u32): Ditto.
32423 (vfmlsl_laneq_low_u32): Ditto.
32424 (vfmlalq_lane_low_u32): Ditto.
32425 (vfmlslq_lane_low_u32): Ditto.
32426 (vfmlalq_laneq_low_u32): Ditto.
32427 (vfmlslq_laneq_low_u32): Ditto.
32428 (vfmlal_lane_high_u32): Ditto.
32429 (vfmlsl_lane_high_u32): Ditto.
32430 (vfmlal_laneq_high_u32): Ditto.
32431 (vfmlsl_laneq_high_u32): Ditto.
32432 (vfmlalq_lane_high_u32): Ditto.
32433 (vfmlslq_lane_high_u32): Ditto.
32434 (vfmlalq_laneq_high_u32): Ditto.
32435 (vfmlslq_laneq_high_u32): Ditto.
32436 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
32437 (AARCH64_FL_FOR_ARCH8_4): New.
32438 (AARCH64_ISA_F16FML): New ISA flag.
32439 (TARGET_F16FML): New feature flag for fp16fml.
32440 (doc/invoke.texi): Document new fp16fml option.
32442 2018-01-10 Michael Collison <michael.collison@arm.com>
32444 * config/aarch64/aarch64-builtins.c:
32445 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
32446 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
32447 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
32448 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
32449 (AARCH64_ISA_SHA3): New ISA flag.
32450 (TARGET_SHA3): New feature flag for sha3.
32451 * config/aarch64/iterators.md (sha512_op): New int attribute.
32452 (CRYPTO_SHA512): New int iterator.
32453 (UNSPEC_SHA512H): New unspec.
32454 (UNSPEC_SHA512H2): Ditto.
32455 (UNSPEC_SHA512SU0): Ditto.
32456 (UNSPEC_SHA512SU1): Ditto.
32457 * config/aarch64/aarch64-simd-builtins.def
32458 (aarch64_crypto_sha512hqv2di): New builtin.
32459 (aarch64_crypto_sha512h2qv2di): Ditto.
32460 (aarch64_crypto_sha512su0qv2di): Ditto.
32461 (aarch64_crypto_sha512su1qv2di): Ditto.
32462 (aarch64_eor3qv8hi): Ditto.
32463 (aarch64_rax1qv2di): Ditto.
32464 (aarch64_xarqv2di): Ditto.
32465 (aarch64_bcaxqv8hi): Ditto.
32466 * config/aarch64/aarch64-simd.md:
32467 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
32468 (aarch64_crypto_sha512su0qv2di): Ditto.
32469 (aarch64_crypto_sha512su1qv2di): Ditto.
32470 (aarch64_eor3qv8hi): Ditto.
32471 (aarch64_rax1qv2di): Ditto.
32472 (aarch64_xarqv2di): Ditto.
32473 (aarch64_bcaxqv8hi): Ditto.
32474 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
32475 (vsha512h2q_u64): Ditto.
32476 (vsha512su0q_u64): Ditto.
32477 (vsha512su1q_u64): Ditto.
32478 (veor3q_u16): Ditto.
32479 (vrax1q_u64): Ditto.
32480 (vxarq_u64): Ditto.
32481 (vbcaxq_u16): Ditto.
32482 * config/arm/types.md (crypto_sha512): New type attribute.
32483 (crypto_sha3): Ditto.
32484 (doc/invoke.texi): Document new sha3 option.
32486 2018-01-10 Michael Collison <michael.collison@arm.com>
32488 * config/aarch64/aarch64-builtins.c:
32489 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
32490 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
32491 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
32492 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
32493 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
32494 (AARCH64_ISA_SM4): New ISA flag.
32495 (TARGET_SM4): New feature flag for sm4.
32496 * config/aarch64/aarch64-simd-builtins.def
32497 (aarch64_sm3ss1qv4si): Ditto.
32498 (aarch64_sm3tt1aq4si): Ditto.
32499 (aarch64_sm3tt1bq4si): Ditto.
32500 (aarch64_sm3tt2aq4si): Ditto.
32501 (aarch64_sm3tt2bq4si): Ditto.
32502 (aarch64_sm3partw1qv4si): Ditto.
32503 (aarch64_sm3partw2qv4si): Ditto.
32504 (aarch64_sm4eqv4si): Ditto.
32505 (aarch64_sm4ekeyqv4si): Ditto.
32506 * config/aarch64/aarch64-simd.md:
32507 (aarch64_sm3ss1qv4si): Ditto.
32508 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
32509 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
32510 (aarch64_sm4eqv4si): Ditto.
32511 (aarch64_sm4ekeyqv4si): Ditto.
32512 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
32513 (sm3part_op): Ditto.
32514 (CRYPTO_SM3TT): Ditto.
32515 (CRYPTO_SM3PART): Ditto.
32516 (UNSPEC_SM3SS1): New unspec.
32517 (UNSPEC_SM3TT1A): Ditto.
32518 (UNSPEC_SM3TT1B): Ditto.
32519 (UNSPEC_SM3TT2A): Ditto.
32520 (UNSPEC_SM3TT2B): Ditto.
32521 (UNSPEC_SM3PARTW1): Ditto.
32522 (UNSPEC_SM3PARTW2): Ditto.
32523 (UNSPEC_SM4E): Ditto.
32524 (UNSPEC_SM4EKEY): Ditto.
32525 * config/aarch64/constraints.md (Ui2): New constraint.
32526 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
32527 * config/arm/types.md (crypto_sm3): New type attribute.
32528 (crypto_sm4): Ditto.
32529 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
32530 (vsm3tt1aq_u32): Ditto.
32531 (vsm3tt1bq_u32): Ditto.
32532 (vsm3tt2aq_u32): Ditto.
32533 (vsm3tt2bq_u32): Ditto.
32534 (vsm3partw1q_u32): Ditto.
32535 (vsm3partw2q_u32): Ditto.
32536 (vsm4eq_u32): Ditto.
32537 (vsm4ekeyq_u32): Ditto.
32538 (doc/invoke.texi): Document new sm4 option.
32540 2018-01-10 Michael Collison <michael.collison@arm.com>
32542 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
32543 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
32544 (AARCH64_FL_FOR_ARCH8_4): New.
32545 (AARCH64_FL_V8_4): New flag.
32546 (doc/invoke.texi): Document new armv8.4-a option.
32548 2018-01-10 Michael Collison <michael.collison@arm.com>
32550 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
32551 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
32552 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
32553 * config/aarch64/aarch64-option-extension.def: Add
32554 AARCH64_OPT_EXTENSION of 'sha2'.
32555 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
32556 (crypto): Disable sha2 and aes if crypto disabled.
32557 (crypto): Enable aes and sha2 if enabled.
32558 (simd): Disable sha2 and aes if simd disabled.
32559 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
32561 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
32562 (TARGET_SHA2): New feature flag for sha2.
32563 (TARGET_AES): New feature flag for aes.
32564 * config/aarch64/aarch64-simd.md:
32565 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
32566 conditional on TARGET_AES.
32567 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
32568 (aarch64_crypto_sha1hsi): Make pattern conditional
32570 (aarch64_crypto_sha1hv4si): Ditto.
32571 (aarch64_be_crypto_sha1hv4si): Ditto.
32572 (aarch64_crypto_sha1su1v4si): Ditto.
32573 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
32574 (aarch64_crypto_sha1su0v4si): Ditto.
32575 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
32576 (aarch64_crypto_sha256su0v4si): Ditto.
32577 (aarch64_crypto_sha256su1v4si): Ditto.
32578 (doc/invoke.texi): Document new aes and sha2 options.
32580 2018-01-10 Martin Sebor <msebor@redhat.com>
32582 PR tree-optimization/83781
32583 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
32586 2018-01-11 Martin Sebor <msebor@gmail.com>
32587 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32589 PR tree-optimization/83501
32590 PR tree-optimization/81703
32592 * tree-ssa-strlen.c (get_string_cst): Rename...
32593 (get_string_len): ...to this. Handle global constants.
32594 (handle_char_store): Adjust.
32596 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
32597 Jim Wilson <jimw@sifive.com>
32599 * config/riscv/riscv-protos.h (riscv_output_return): New.
32600 * config/riscv/riscv.c (struct machine_function): New naked_p field.
32601 (riscv_attribute_table, riscv_output_return),
32602 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
32603 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
32604 (riscv_compute_frame_info): Only compute frame->mask if not a naked
32606 (riscv_expand_prologue): Add early return for naked function.
32607 (riscv_expand_epilogue): Likewise.
32608 (riscv_function_ok_for_sibcall): Return false for naked function.
32609 (riscv_set_current_function): New.
32610 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
32611 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
32612 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
32613 * doc/extend.texi (RISC-V Function Attributes): New.
32615 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
32617 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
32618 check for 128-bit long double before checking TCmode.
32619 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
32620 128-bit long doubles before checking TFmode or TCmode.
32621 (FLOAT128_IBM_P): Likewise.
32623 2018-01-10 Martin Sebor <msebor@redhat.com>
32625 PR tree-optimization/83671
32626 * builtins.c (c_strlen): Unconditionally return zero for the empty
32628 Use -Warray-bounds for warnings.
32629 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
32630 for non-constant array indices with COMPONENT_REF, arrays of
32631 arrays, and pointers to arrays.
32632 (gimple_fold_builtin_strlen): Determine and set length range for
32633 non-constant character arrays.
32635 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
32637 PR middle-end/81897
32638 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
32641 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
32643 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
32645 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
32648 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
32649 VECTOR_MEM_ALTIVEC_OR_VSX_P.
32650 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
32651 indexed_or_indirect_operand predicate.
32652 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
32653 (*vsx_le_perm_load_v8hi): Likewise.
32654 (*vsx_le_perm_load_v16qi): Likewise.
32655 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
32656 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
32657 (*vsx_le_perm_store_v8hi): Likewise.
32658 (*vsx_le_perm_store_v16qi): Likewise.
32659 (eight unnamed splitters): Likewise.
32661 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
32663 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
32664 * config/rs6000/emmintrin.h: Likewise.
32665 * config/rs6000/mmintrin.h: Likewise.
32666 * config/rs6000/xmmintrin.h: Likewise.
32668 2018-01-10 David Malcolm <dmalcolm@redhat.com>
32671 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
32673 * tree.c (tree_nop_conversion): Return true for location wrapper
32675 (maybe_wrap_with_location): New function.
32676 (selftest::check_strip_nops): New function.
32677 (selftest::test_location_wrappers): New function.
32678 (selftest::tree_c_tests): Call it.
32679 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
32680 (maybe_wrap_with_location): New decl.
32681 (EXPR_LOCATION_WRAPPER_P): New macro.
32682 (location_wrapper_p): New inline function.
32683 (tree_strip_any_location_wrapper): New inline function.
32685 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
32688 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
32689 stack_realign_offset for the largest alignment of stack slot
32691 (ix86_find_max_used_stack_alignment): New function.
32692 (ix86_finalize_stack_frame_flags): Use it. Set
32693 max_used_stack_alignment if we don't realign stack.
32694 * config/i386/i386.h (machine_function): Add
32695 max_used_stack_alignment.
32697 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
32699 * config/arm/arm.opt (-mbranch-cost): New option.
32700 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
32703 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
32706 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
32707 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
32709 2018-01-10 Richard Biener <rguenther@suse.de>
32712 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
32713 early out so it also covers the case where we have a non-NULL
32716 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
32718 PR tree-optimization/83753
32719 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
32720 for non-strided grouped accesses if the number of elements is 1.
32722 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
32725 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
32726 * i386.h (TARGET_USE_GATHER): Define.
32727 * x86-tune.def (X86_TUNE_USE_GATHER): New.
32729 2018-01-10 Martin Liska <mliska@suse.cz>
32732 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
32733 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
32735 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
32736 CLEANUP_NO_PARTITIONING is not set.
32738 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
32740 * doc/rtl.texi: Remove documentation of (const ...) wrappers
32741 for vectors, as a partial revert of r254296.
32742 * rtl.h (const_vec_p): Delete.
32743 (const_vec_duplicate_p): Don't test for vector CONSTs.
32744 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
32745 * expmed.c (make_tree): Likewise.
32748 * common.md (E, F): Use CONSTANT_P instead of checking for
32750 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
32751 checking for CONST_VECTOR.
32753 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
32755 PR middle-end/83575
32756 * predict.c (force_edge_cold): Handle in more sane way edges
32757 with no prediction.
32759 2018-01-09 Carl Love <cel@us.ibm.com>
32761 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
32763 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
32764 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
32765 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
32766 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
32767 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
32768 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
32769 * config/rs6000/rs6000-protos.h: Add extern defition for
32770 rs6000_generate_float2_double_code.
32771 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
32773 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
32774 (float2_v2df): Add define_expand.
32776 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
32779 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
32780 op_mode in the force_to_mode call.
32782 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
32784 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
32785 instead of checking each element individually.
32786 (aarch64_evpc_uzp): Likewise.
32787 (aarch64_evpc_zip): Likewise.
32788 (aarch64_evpc_ext): Likewise.
32789 (aarch64_evpc_rev): Likewise.
32790 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
32791 instead of checking each element individually. Return true without
32793 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
32794 whether all selected elements come from the same input, instead of
32795 checking each element individually. Remove calls to gen_rtx_REG,
32796 start_sequence and end_sequence and instead assert that no rtl is
32799 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
32801 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
32802 order of HIGH and CONST checks.
32804 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
32806 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
32807 if the destination isn't an SSA_NAME.
32809 2018-01-09 Richard Biener <rguenther@suse.de>
32811 PR tree-optimization/83668
32812 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
32814 (canonicalize_loop_form): ... here, renamed from ...
32815 (canonicalize_loop_closed_ssa_form): ... this and amended to
32816 swap successor edges for loop exit blocks to make us use
32817 the RPO order we need for initial schedule generation.
32819 2018-01-09 Joseph Myers <joseph@codesourcery.com>
32821 PR tree-optimization/64811
32822 * match.pd: When optimizing comparisons with Inf, avoid
32823 introducing or losing exceptions from comparisons with NaN.
32825 2018-01-09 Martin Liska <mliska@suse.cz>
32828 * asan.c (shadow_mem_size): Add gcc_assert.
32830 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
32832 Don't save registers in main().
32835 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
32836 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
32837 * config/avr/avr.c (avr_set_current_function): Don't error if
32838 naked, OS_task or OS_main are specified at the same time.
32839 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
32841 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
32843 * common/config/avr/avr-common.c (avr_option_optimization_table):
32844 Switch on -mmain-is-OS_task for optimizing compilations.
32846 2018-01-09 Richard Biener <rguenther@suse.de>
32848 PR tree-optimization/83572
32849 * graphite.c: Include cfganal.h.
32850 (graphite_transform_loops): Connect infinite loops to exit
32851 and remove fake edges at the end.
32853 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
32855 * ipa-inline.c (edge_badness): Revert accidental checkin.
32857 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
32860 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
32861 symbols; not inline clones.
32863 2018-01-09 Jakub Jelinek <jakub@redhat.com>
32866 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
32867 hard registers. Formatting fixes.
32869 PR preprocessor/83722
32870 * gcc.c (try_generate_repro): Pass
32871 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
32872 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
32875 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
32876 Kito Cheng <kito.cheng@gmail.com>
32878 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
32879 (riscv_leaf_function_p): Delete.
32880 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
32882 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32884 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
32886 (do_ifelse): New function.
32887 (do_isel): New function.
32888 (do_sub3): New function.
32889 (do_add3): New function.
32890 (do_load_mask_compare): New function.
32891 (do_overlap_load_compare): New function.
32892 (expand_compare_loop): New function.
32893 (expand_block_compare): Call expand_compare_loop() when appropriate.
32894 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
32895 option description.
32896 (-mblock-compare-inline-loop-limit): New option.
32898 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32901 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
32902 Reverse order of second and third operands in first alternative.
32903 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
32904 of first and second elements in UNSPEC_VPERMR vector.
32905 (altivec_expand_vec_perm_le): Likewise.
32907 2018-01-08 Jeff Law <law@redhat.com>
32909 PR rtl-optimizatin/81308
32910 * tree-switch-conversion.c (cfg_altered): New file scoped static.
32911 (process_switch): If group_case_labels makes a change, then set
32913 (pass_convert_switch::execute): If a switch is converted, then
32914 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
32916 PR rtl-optimization/81308
32917 * recog.c (split_all_insns): Conditionally cleanup the CFG after
32920 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
32922 PR target/83663 - Revert r255946
32923 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
32924 generation for cases where splatting a value is not useful.
32925 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
32926 across a vec_duplicate and a paradoxical subreg forming a vector
32927 mode to a vec_concat.
32929 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32931 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
32932 -march=armv8.3-a variants.
32933 * config/arm/t-multilib: Likewise.
32934 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
32936 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32938 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
32940 (cceq_ior_compare_complement): Give it a name so I can use it, and
32941 change boolean_or_operator predicate to boolean_operator so it can
32942 be used to generate a crand.
32943 (eqne): New code iterator.
32944 (bd/bd_neg): New code_attrs.
32945 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
32946 a single define_insn.
32947 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
32948 decrement (bdnzt/bdnzf/bdzt/bdzf).
32949 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
32950 with the new names of the branch decrement patterns, and added the
32951 names of the branch decrement conditional patterns.
32953 2018-01-08 Richard Biener <rguenther@suse.de>
32955 PR tree-optimization/83563
32956 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
32959 2018-01-08 Richard Biener <rguenther@suse.de>
32961 PR middle-end/83713
32962 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
32964 2018-01-08 Richard Biener <rguenther@suse.de>
32966 PR tree-optimization/83685
32967 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
32968 references to abnormals.
32970 2018-01-08 Richard Biener <rguenther@suse.de>
32973 * dwarf2out.c (output_indirect_strings): Handle empty
32974 skeleton_debug_str_hash.
32975 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
32977 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
32979 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
32980 (emit_store_direct): Likewise.
32981 (arc_trampoline_adjust_address): Likewise.
32982 (arc_asm_trampoline_template): New function.
32983 (arc_initialize_trampoline): Use asm_trampoline_template.
32984 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
32985 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
32986 * config/arc/arc.md (flush_icache): Delete pattern.
32988 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
32990 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
32991 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
32994 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
32997 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
32998 by not USED_FOR_TARGET.
32999 (make_pass_resolve_sw_modes): Likewise.
33001 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
33003 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
33006 2018-01-08 Richard Biener <rguenther@suse.de>
33008 PR middle-end/83580
33009 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
33011 2018-01-08 Richard Biener <rguenther@suse.de>
33013 PR middle-end/83517
33014 * match.pd ((t * 2) / 2) -> t): Add missing :c.
33016 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
33018 PR middle-end/81897
33019 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
33020 basic blocks with a small number of successors.
33021 (convert_control_dep_chain_into_preds): Improve handling of
33023 (dump_predicates): Split apart into...
33024 (dump_pred_chain): ...here...
33025 (dump_pred_info): ...and here.
33026 (can_one_predicate_be_invalidated_p): Add debugging printfs.
33027 (can_chain_union_be_invalidated_p): Improve check for invalidation
33029 (uninit_uses_cannot_happen): Avoid unnecessary if
33030 convert_control_dep_chain_into_preds yielded nothing.
33032 2018-01-06 Martin Sebor <msebor@redhat.com>
33034 PR tree-optimization/83640
33035 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
33036 subtracting negative offset from size.
33037 (builtin_access::overlap): Adjust offset bounds of the access to fall
33038 within the size of the object if possible.
33040 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
33042 PR rtl-optimization/83699
33043 * expmed.c (extract_bit_field_1): Restrict the vector usage of
33044 extract_bit_field_as_subreg to cases in which the extracted
33045 value is also a vector.
33047 * lra-constraints.c (process_alt_operands): Test for the equivalence
33048 substitutions when detecting a possible reload cycle.
33050 2018-01-06 Jakub Jelinek <jakub@redhat.com>
33053 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
33054 by default if flag_selective_schedling{,2}. Formatting fixes.
33056 PR rtl-optimization/83682
33057 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
33058 if it has non-VECTOR_MODE element mode.
33059 (vec_duplicate_p): Likewise.
33061 PR middle-end/83694
33062 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
33063 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
33065 2018-01-05 Jakub Jelinek <jakub@redhat.com>
33068 * config/i386/i386-builtin.def
33069 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
33070 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
33071 Require also OPTION_MASK_ISA_AVX512F in addition to
33072 OPTION_MASK_ISA_GFNI.
33073 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
33074 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
33075 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
33076 to OPTION_MASK_ISA_GFNI.
33077 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
33078 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
33079 OPTION_MASK_ISA_AVX512BW.
33080 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
33081 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
33082 addition to OPTION_MASK_ISA_GFNI.
33083 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
33084 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
33085 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
33086 to OPTION_MASK_ISA_GFNI.
33087 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
33088 a requirement for all ISAs rather than any of them with a few
33090 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
33092 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
33093 bitmasks to be enabled with 3 exceptions, instead of requiring any
33094 enabled ISA with lots of exceptions.
33095 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
33096 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
33097 Change avx512bw in isa attribute to avx512f.
33098 * config/i386/sgxintrin.h: Add license boilerplate.
33099 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
33100 to __AVX512F__ and __AVX512VL to __AVX512VL__.
33101 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
33102 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
33104 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
33105 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
33106 temporarily sse2 rather than sse if not enabled already.
33109 * config/i386/sse.md (VI248_VLBW): Rename to ...
33110 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
33111 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
33112 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
33113 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
33114 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
33115 mode iterator instead of VI248_VLBW.
33117 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
33119 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
33120 (record_modified): Skip clobbers; add debug output.
33121 (param_change_prob): Use sreal frequencies.
33123 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
33125 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
33126 punt for user-aligned variables.
33128 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
33130 * tree-chrec.c (chrec_contains_symbols): Return true for
33133 2018-01-05 Sudakshina Das <sudi.das@arm.com>
33136 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
33137 of (x|y) == x for BICS pattern.
33139 2018-01-05 Jakub Jelinek <jakub@redhat.com>
33141 PR tree-optimization/83605
33142 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
33143 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
33146 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
33148 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
33149 * config/epiphany/rtems.h: New file.
33151 2018-01-04 Jakub Jelinek <jakub@redhat.com>
33152 Uros Bizjak <ubizjak@gmail.com>
33155 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
33156 QIreg_operand instead of register_operand predicate.
33157 * config/i386/i386.c (ix86_rop_should_change_byte_p,
33158 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
33159 comments instead of -fmitigate[-_]rop.
33161 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33164 * cgraphunit.c (symbol_table::compile): Switch to text_section
33165 before calling assembly_start debug hook.
33166 * run-rtl-passes.c (run_rtl_passes): Likewise.
33169 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33171 * tree-vrp.c (extract_range_from_binary_expr_1): Check
33172 range_int_cst_p rather than !symbolic_range_p before calling
33173 extract_range_from_multiplicative_op_1.
33175 2018-01-04 Jeff Law <law@redhat.com>
33177 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
33178 redundant test in assertion.
33180 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33182 * doc/rtl.texi: Document machine_mode wrapper classes.
33184 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33186 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
33187 using tree_to_uhwi.
33189 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33191 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
33192 the VEC_PERM_EXPR fold to fail.
33194 2018-01-04 Jakub Jelinek <jakub@redhat.com>
33197 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
33198 to switched_sections.
33200 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33203 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
33204 test for d.testing.
33206 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
33209 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
33210 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
33212 2018-01-04 Jakub Jelinek <jakub@redhat.com>
33215 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
33216 is BLKmode and bitpos not zero or mode change is needed.
33218 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33221 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
33224 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
33227 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
33228 instead of MULT rtx. Update all corresponding splitters.
33229 (*saddl_se): Ditto.
33230 (*ssub<modesuffix>): Ditto.
33231 (*ssubl_se): Ditto.
33232 (*cmp_sadd_di): Update split patterns.
33233 (*cmp_sadd_si): Ditto.
33234 (*cmp_sadd_sidi): Ditto.
33235 (*cmp_ssub_di): Ditto.
33236 (*cmp_ssub_si): Ditto.
33237 (*cmp_ssub_sidi): Ditto.
33238 * config/alpha/predicates.md (const23_operand): New predicate.
33239 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
33240 Look for ASHIFT, not MULT inner operand.
33241 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
33243 2018-01-04 Martin Liska <mliska@suse.cz>
33245 PR gcov-profile/83669
33246 * gcov.c (output_intermediate_file): Add version to intermediate
33248 * doc/gcov.texi: Document new field 'version' in intermediate
33249 file format. Fix location of '-k' option of gcov command.
33251 2018-01-04 Martin Liska <mliska@suse.cz>
33254 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
33256 2018-01-04 Jakub Jelinek <jakub@redhat.com>
33258 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
33260 2018-01-03 Martin Sebor <msebor@redhat.com>
33262 PR tree-optimization/83655
33263 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
33264 checking calls with invalid arguments.
33266 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33268 * tree-vect-stmts.c (vect_get_store_rhs): New function.
33269 (vectorizable_mask_load_store): Delete.
33270 (vectorizable_call): Return false for masked loads and stores.
33271 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
33272 instead of gimple_assign_rhs1.
33273 (vectorizable_load): Handle IFN_MASK_LOAD.
33274 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
33276 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33278 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
33280 (vectorizable_mask_load_store): ...here.
33281 (vectorizable_load): ...and here.
33283 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33285 * tree-vect-stmts.c (vect_build_all_ones_mask)
33286 (vect_build_zero_merge_argument): New functions, split out from...
33287 (vectorizable_load): ...here.
33289 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33291 * tree-vect-stmts.c (vect_check_store_rhs): New function,
33293 (vectorizable_mask_load_store): ...here.
33294 (vectorizable_store): ...and here.
33296 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33298 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
33300 (vectorizable_mask_load_store): ...here.
33302 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33304 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
33305 (vect_model_store_cost): Take a vec_load_store_type instead of a
33307 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
33308 (vect_model_store_cost): Take a vec_load_store_type instead of a
33310 (vectorizable_mask_load_store): Update accordingly.
33311 (vectorizable_store): Likewise.
33312 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
33314 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33316 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
33317 IFN_MASK_LOAD calls here rather than...
33318 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
33320 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33321 Alan Hayward <alan.hayward@arm.com>
33322 David Sherwood <david.sherwood@arm.com>
33324 * expmed.c (extract_bit_field_1): For vector extracts,
33325 fall back to extract_bit_field_as_subreg if vec_extract
33328 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33329 Alan Hayward <alan.hayward@arm.com>
33330 David Sherwood <david.sherwood@arm.com>
33332 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
33333 they are variable or constant sized.
33334 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
33335 slots for constant-sized data.
33337 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33338 Alan Hayward <alan.hayward@arm.com>
33339 David Sherwood <david.sherwood@arm.com>
33341 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
33342 handling COND_EXPRs with boolean comparisons, try to find a better
33343 basis for the mask type than the boolean itself.
33345 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33347 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
33348 is calculated and how it can be overridden.
33349 * genmodes.c (max_bitsize_mode_any_mode): New variable.
33350 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
33352 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
33355 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33356 Alan Hayward <alan.hayward@arm.com>
33357 David Sherwood <david.sherwood@arm.com>
33359 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
33360 Remove the mode argument.
33361 (aarch64_simd_valid_immediate): Remove the mode and inverse
33363 * config/aarch64/iterators.md (bitsize): New iterator.
33364 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
33365 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
33366 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
33367 aarch64_simd_valid_immediate.
33368 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
33369 (aarch64_reg_or_bic_imm): Likewise.
33370 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
33371 with an insn_type enum and msl with a modifier_type enum.
33372 Replace element_width with a scalar_mode. Change the shift
33373 to unsigned int. Add constructors for scalar_float_mode and
33374 scalar_int_mode elements.
33375 (aarch64_vect_float_const_representable_p): Delete.
33376 (aarch64_can_const_movi_rtx_p)
33377 (aarch64_simd_scalar_immediate_valid_for_move)
33378 (aarch64_simd_make_constant): Update call to
33379 aarch64_simd_valid_immediate.
33380 (aarch64_advsimd_valid_immediate_hs): New function.
33381 (aarch64_advsimd_valid_immediate): Likewise.
33382 (aarch64_simd_valid_immediate): Remove mode and inverse
33383 arguments. Rewrite to use the above. Use const_vec_duplicate_p
33384 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
33385 and aarch64_float_const_representable_p on the result.
33386 (aarch64_output_simd_mov_immediate): Remove mode argument.
33387 Update call to aarch64_simd_valid_immediate and use of
33388 simd_immediate_info.
33389 (aarch64_output_scalar_simd_mov_immediate): Update call
33392 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33393 Alan Hayward <alan.hayward@arm.com>
33394 David Sherwood <david.sherwood@arm.com>
33396 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
33397 (mode_nunits): Likewise CONST_MODE_NUNITS.
33398 * machmode.def (ADJUST_NUNITS): Document.
33399 * genmodes.c (mode_data::need_nunits_adj): New field.
33400 (blank_mode): Update accordingly.
33401 (adj_nunits): New variable.
33402 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
33404 (emit_mode_size_inline): Set need_bytesize_adj for all modes
33405 listed in adj_nunits.
33406 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
33407 listed in adj_nunits. Don't emit case statements for such modes.
33408 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
33409 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
33410 nothing if adj_nunits is nonnull.
33411 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
33412 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
33413 (emit_mode_fbit): Update use of print_maybe_const_decl.
33414 (emit_move_size): Likewise. Treat the array as non-const
33416 (emit_mode_adjustments): Handle adj_nunits.
33418 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33420 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
33421 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
33422 (VECTOR_MODES): Use it.
33423 (make_vector_modes): Take the prefix as an argument.
33425 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33426 Alan Hayward <alan.hayward@arm.com>
33427 David Sherwood <david.sherwood@arm.com>
33429 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
33430 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
33431 for MODE_VECTOR_BOOL.
33432 * machmode.def (VECTOR_BOOL_MODE): Document.
33433 * genmodes.c (VECTOR_BOOL_MODE): New macro.
33434 (make_vector_bool_mode): New function.
33435 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
33437 * lto-streamer-in.c (lto_input_mode_table): Likewise.
33438 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
33440 * stor-layout.c (int_mode_for_mode): Likewise.
33441 * tree.c (build_vector_type_for_mode): Likewise.
33442 * varasm.c (output_constant_pool_2): Likewise.
33443 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
33444 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
33445 for MODE_VECTOR_BOOL.
33446 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
33447 of mode class checks.
33448 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
33449 instead of a list of mode class checks.
33450 (expand_vector_scalar_condition): Likewise.
33451 (type_for_widest_vector_mode): Handle BImode as an inner mode.
33453 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33454 Alan Hayward <alan.hayward@arm.com>
33455 David Sherwood <david.sherwood@arm.com>
33457 * machmode.h (mode_size): Change from unsigned short to
33459 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
33460 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
33461 or if measurement_type is not polynomial.
33462 (fixed_size_mode::includes_p): Check for constant-sized modes.
33463 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
33464 return a poly_uint16 rather than an unsigned short.
33465 (emit_mode_size): Change the type of mode_size from unsigned short
33466 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
33467 (emit_mode_adjustments): Cope with polynomial vector sizes.
33468 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
33470 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
33472 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
33473 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
33474 * caller-save.c (setup_save_areas): Likewise.
33475 (replace_reg_with_saved_mem): Likewise.
33476 * calls.c (emit_library_call_value_1): Likewise.
33477 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
33478 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
33479 (gen_lowpart_for_combine): Likewise.
33480 * convert.c (convert_to_integer_1): Likewise.
33481 * cse.c (equiv_constant, cse_insn): Likewise.
33482 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
33483 (cselib_subst_to_values): Likewise.
33484 * dce.c (word_dce_process_block): Likewise.
33485 * df-problems.c (df_word_lr_mark_ref): Likewise.
33486 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
33487 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
33488 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
33489 (rtl_for_decl_location): Likewise.
33490 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
33491 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
33492 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
33493 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
33494 (expand_expr_real_1): Likewise.
33495 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
33496 (pad_below): Likewise.
33497 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
33498 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
33499 * ira.c (get_subreg_tracking_sizes): Likewise.
33500 * ira-build.c (ira_create_allocno_objects): Likewise.
33501 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
33502 (ira_sort_regnos_for_alter_reg): Likewise.
33503 * ira-costs.c (record_operand_costs): Likewise.
33504 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
33505 (resolve_simple_move): Likewise.
33506 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
33507 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
33508 (lra_constraints): Likewise.
33509 (CONST_POOL_OK_P): Reject variable-sized modes.
33510 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
33511 (add_pseudo_to_slot, lra_spill): Likewise.
33512 * omp-low.c (omp_clause_aligned_alignment): Likewise.
33513 * optabs-query.c (get_best_extraction_insn): Likewise.
33514 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
33515 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
33516 (expand_mult_highpart, valid_multiword_target_p): Likewise.
33517 * recog.c (offsettable_address_addr_space_p): Likewise.
33518 * regcprop.c (maybe_mode_change): Likewise.
33519 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
33520 * regrename.c (build_def_use): Likewise.
33521 * regstat.c (dump_reg_info): Likewise.
33522 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
33523 (find_reloads, find_reloads_subreg_address): Likewise.
33524 * reload1.c (eliminate_regs_1): Likewise.
33525 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
33526 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
33527 (simplify_binary_operation_1, simplify_subreg): Likewise.
33528 * targhooks.c (default_function_arg_padding): Likewise.
33529 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
33530 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
33531 (verify_gimple_assign_ternary): Likewise.
33532 * tree-inline.c (estimate_move_cost): Likewise.
33533 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
33534 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
33535 (get_address_cost_ainc): Likewise.
33536 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33537 (vect_supportable_dr_alignment): Likewise.
33538 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
33539 (vectorizable_reduction): Likewise.
33540 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
33541 (vectorizable_operation, vectorizable_load): Likewise.
33542 * tree.c (build_same_sized_truth_vector_type): Likewise.
33543 * valtrack.c (cleanup_auto_inc_dec): Likewise.
33544 * var-tracking.c (emit_note_insn_var_location): Likewise.
33545 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
33546 (ADDR_VEC_ALIGN): Likewise.
33548 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33549 Alan Hayward <alan.hayward@arm.com>
33550 David Sherwood <david.sherwood@arm.com>
33552 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
33554 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
33555 or if measurement_type is polynomial.
33556 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
33557 * combine.c (make_extraction): Likewise.
33558 * dse.c (find_shift_sequence): Likewise.
33559 * dwarf2out.c (mem_loc_descriptor): Likewise.
33560 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
33561 (extract_bit_field, extract_low_bits): Likewise.
33562 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
33563 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
33564 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
33565 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
33566 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
33567 * reload.c (find_reloads): Likewise.
33568 * reload1.c (alter_reg): Likewise.
33569 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
33570 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
33571 * tree-if-conv.c (predicate_mem_writes): Likewise.
33572 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
33573 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
33574 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
33575 * valtrack.c (dead_debug_insert_temp): Likewise.
33576 * varasm.c (mergeable_constant_section): Likewise.
33577 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
33579 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33580 Alan Hayward <alan.hayward@arm.com>
33581 David Sherwood <david.sherwood@arm.com>
33583 * expr.c (expand_assignment): Cope with polynomial mode sizes
33584 when assigning to a CONCAT.
33586 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33587 Alan Hayward <alan.hayward@arm.com>
33588 David Sherwood <david.sherwood@arm.com>
33590 * machmode.h (mode_precision): Change from unsigned short to
33592 (mode_to_precision): Return a poly_uint16 rather than an unsigned
33594 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
33595 or if measurement_type is not polynomial.
33596 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
33597 in which the mode is already known to be a scalar_int_mode.
33598 * genmodes.c (emit_mode_precision): Change the type of mode_precision
33599 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
33601 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
33602 for GET_MODE_PRECISION.
33603 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
33604 for GET_MODE_PRECISION.
33605 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
33607 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
33608 (expand_field_assignment, make_extraction): Likewise.
33609 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
33610 (get_last_value): Likewise.
33611 * convert.c (convert_to_integer_1): Likewise.
33612 * cse.c (cse_insn): Likewise.
33613 * expr.c (expand_expr_real_1): Likewise.
33614 * lra-constraints.c (simplify_operand_subreg): Likewise.
33615 * optabs-query.c (can_atomic_load_p): Likewise.
33616 * optabs.c (expand_atomic_load): Likewise.
33617 (expand_atomic_store): Likewise.
33618 * ree.c (combine_reaching_defs): Likewise.
33619 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
33620 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
33621 * tree.h (type_has_mode_precision_p): Likewise.
33622 * ubsan.c (instrument_si_overflow): Likewise.
33624 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33625 Alan Hayward <alan.hayward@arm.com>
33626 David Sherwood <david.sherwood@arm.com>
33628 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
33629 polynomial numbers of units.
33630 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
33631 (valid_vector_subparts_p): New function.
33632 (build_vector_type): Remove temporary shim and take the number
33633 of units as a poly_uint64 rather than an int.
33634 (build_opaque_vector_type): Take the number of units as a
33635 poly_uint64 rather than an int.
33636 * tree.c (build_vector_from_ctor): Handle polynomial
33637 TYPE_VECTOR_SUBPARTS.
33638 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
33639 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
33640 (build_vector_from_val): If the number of units is variable,
33641 use build_vec_duplicate_cst for constant operands and
33642 VEC_DUPLICATE_EXPR otherwise.
33643 (make_vector_type): Remove temporary is_constant ().
33644 (build_vector_type, build_opaque_vector_type): Take the number of
33645 units as a poly_uint64 rather than an int.
33646 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
33648 * cfgexpand.c (expand_debug_expr): Likewise.
33649 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
33650 (store_constructor, expand_expr_real_1): Likewise.
33651 (const_scalar_mask_from_tree): Likewise.
33652 * fold-const-call.c (fold_const_reduction): Likewise.
33653 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
33654 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
33655 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
33656 (fold_relational_const): Likewise.
33657 (native_interpret_vector): Likewise. Change the size from an
33658 int to an unsigned int.
33659 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
33660 TYPE_VECTOR_SUBPARTS.
33661 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
33662 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
33663 duplicating a non-constant operand into a variable-length vector.
33664 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
33665 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
33666 * ipa-icf.c (sem_variable::equals): Likewise.
33667 * match.pd: Likewise.
33668 * omp-simd-clone.c (simd_clone_subparts): Likewise.
33669 * print-tree.c (print_node): Likewise.
33670 * stor-layout.c (layout_type): Likewise.
33671 * targhooks.c (default_builtin_vectorization_cost): Likewise.
33672 * tree-cfg.c (verify_gimple_comparison): Likewise.
33673 (verify_gimple_assign_binary): Likewise.
33674 (verify_gimple_assign_ternary): Likewise.
33675 (verify_gimple_assign_single): Likewise.
33676 * tree-pretty-print.c (dump_generic_node): Likewise.
33677 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
33678 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
33679 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
33680 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
33681 (vect_shift_permute_load_chain): Likewise.
33682 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
33683 (expand_vector_condition, optimize_vector_constructor): Likewise.
33684 (lower_vec_perm, get_compute_type): Likewise.
33685 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
33686 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
33687 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
33688 (vect_recog_mask_conversion_pattern): Likewise.
33689 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
33690 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
33691 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
33692 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
33693 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
33694 (vectorizable_shift, vectorizable_operation, vectorizable_store)
33695 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
33696 (supportable_widening_operation): Likewise.
33697 (supportable_narrowing_operation): Likewise.
33698 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
33700 * varasm.c (output_constant): Likewise.
33702 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33703 Alan Hayward <alan.hayward@arm.com>
33704 David Sherwood <david.sherwood@arm.com>
33706 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
33707 so that both the length == 3 and length != 3 cases set up their
33708 own permute vectors. Add comments explaining why we know the
33709 number of elements is constant.
33710 (vect_permute_load_chain): Likewise.
33712 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33713 Alan Hayward <alan.hayward@arm.com>
33714 David Sherwood <david.sherwood@arm.com>
33716 * machmode.h (mode_nunits): Change from unsigned char to
33718 (ONLY_FIXED_SIZE_MODES): New macro.
33719 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
33720 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
33721 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
33723 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
33724 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
33725 or if measurement_type is not polynomial.
33726 * genmodes.c (ZERO_COEFFS): New macro.
33727 (emit_mode_nunits_inline): Make mode_nunits_inline return a
33729 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
33730 Use ZERO_COEFFS when emitting initializers.
33731 * data-streamer.h (bp_pack_poly_value): New function.
33732 (bp_unpack_poly_value): Likewise.
33733 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
33734 for GET_MODE_NUNITS.
33735 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
33736 for GET_MODE_NUNITS.
33737 * tree.c (make_vector_type): Remove temporary shim and make
33738 the real function take the number of units as a poly_uint64
33739 rather than an int.
33740 (build_vector_type_for_mode): Handle polynomial nunits.
33741 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
33742 * emit-rtl.c (const_vec_series_p_1): Likewise.
33743 (gen_rtx_CONST_VECTOR): Likewise.
33744 * fold-const.c (test_vec_duplicate_folding): Likewise.
33745 * genrecog.c (validate_pattern): Likewise.
33746 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
33747 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
33748 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
33749 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
33750 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
33751 * rtlanal.c (subreg_get_info): Likewise.
33752 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
33753 (vect_grouped_load_supported): Likewise.
33754 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
33755 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
33756 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
33757 (simplify_const_unary_operation, simplify_binary_operation_1)
33758 (simplify_const_binary_operation, simplify_ternary_operation)
33759 (test_vector_ops_duplicate, test_vector_ops): Likewise.
33760 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
33761 instead of CONST_VECTOR_NUNITS.
33762 * varasm.c (output_constant_pool_2): Likewise.
33763 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
33764 explicit-encoded elements in the XVEC for variable-length vectors.
33766 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33768 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
33770 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33771 Alan Hayward <alan.hayward@arm.com>
33772 David Sherwood <david.sherwood@arm.com>
33774 * coretypes.h (fixed_size_mode): Declare.
33775 (fixed_size_mode_pod): New typedef.
33776 * builtins.h (target_builtins::x_apply_args_mode)
33777 (target_builtins::x_apply_result_mode): Change type to
33778 fixed_size_mode_pod.
33779 * builtins.c (apply_args_size, apply_result_size, result_vector)
33780 (expand_builtin_apply_args_1, expand_builtin_apply)
33781 (expand_builtin_return): Update accordingly.
33783 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33785 * cse.c (hash_rtx_cb): Hash only the encoded elements.
33786 * cselib.c (cselib_hash_rtx): Likewise.
33787 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
33788 CONST_VECTOR encoding.
33790 2018-01-03 Jakub Jelinek <jakub@redhat.com>
33791 Jeff Law <law@redhat.com>
33794 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
33795 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
33796 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
33797 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
33800 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
33801 explicitly probe *sp in a noreturn function if there were any callee
33802 register saves or frame pointer is needed.
33804 2018-01-03 Jakub Jelinek <jakub@redhat.com>
33807 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
33808 BLKmode for ternary, binary or unary expressions.
33811 * var-tracking.c (delete_vta_debug_insn): New inline function.
33812 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
33813 insns from get_insns () to NULL instead of each bb separately.
33814 Use delete_vta_debug_insn. No longer static.
33815 (vt_debug_insns_local, variable_tracking_main_1): Adjust
33816 delete_vta_debug_insns callers.
33817 * rtl.h (delete_vta_debug_insns): Declare.
33818 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
33819 instead of variable_tracking_main.
33821 2018-01-03 Martin Sebor <msebor@redhat.com>
33823 PR tree-optimization/83603
33824 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
33825 arguments past the endof the argument list in functions declared
33826 without a prototype.
33827 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
33828 Avoid checking when arguments are null.
33830 2018-01-03 Martin Sebor <msebor@redhat.com>
33833 * doc/extend.texi (attribute const): Fix a typo.
33834 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
33835 issuing -Wsuggest-attribute for void functions.
33837 2018-01-03 Martin Sebor <msebor@redhat.com>
33839 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
33840 offset_int::from instead of wide_int::to_shwi.
33841 (maybe_diag_overlap): Remove assertion.
33842 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
33843 * gimple-ssa-sprintf.c (format_directive): Same.
33844 (parse_directive): Same.
33845 (sprintf_dom_walker::compute_format_length): Same.
33846 (try_substitute_return_value): Same.
33848 2018-01-03 Jeff Law <law@redhat.com>
33850 PR middle-end/83654
33851 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
33852 non-constant residual for zero at runtime and avoid probing in
33853 that case. Reorganize code for trailing problem to mirror handling
33856 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33858 PR tree-optimization/83501
33859 * tree-ssa-strlen.c (get_string_cst): New.
33860 (handle_char_store): Call get_string_cst.
33862 2018-01-03 Martin Liska <mliska@suse.cz>
33864 PR tree-optimization/83593
33865 * tree-ssa-strlen.c: Include tree-cfg.h.
33866 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
33867 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
33868 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
33870 (strlen_dom_walker::before_dom_children): Call
33871 gimple_purge_dead_eh_edges. Dump tranformation with details
33873 (strlen_dom_walker::before_dom_children): Update call by adding
33874 new argument cleanup_eh.
33875 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
33877 2018-01-03 Martin Liska <mliska@suse.cz>
33880 * cif-code.def (VARIADIC_THUNK): New enum value.
33881 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
33884 2018-01-03 Jan Beulich <jbeulich@suse.com>
33886 * sse.md (mov<mode>_internal): Tighten condition for when to use
33887 vmovdqu<ssescalarsize> for TI and OI modes.
33889 2018-01-03 Jakub Jelinek <jakub@redhat.com>
33891 Update copyright years.
33893 2018-01-03 Martin Liska <mliska@suse.cz>
33896 * ipa-visibility.c (function_and_variable_visibility): Skip
33897 functions with noipa attribure.
33899 2018-01-03 Jakub Jelinek <jakub@redhat.com>
33901 * gcc.c (process_command): Update copyright notice dates.
33902 * gcov-dump.c (print_version): Ditto.
33903 * gcov.c (print_version): Ditto.
33904 * gcov-tool.c (print_version): Ditto.
33905 * gengtype.c (create_file): Ditto.
33906 * doc/cpp.texi: Bump @copying's copyright year.
33907 * doc/cppinternals.texi: Ditto.
33908 * doc/gcc.texi: Ditto.
33909 * doc/gccint.texi: Ditto.
33910 * doc/gcov.texi: Ditto.
33911 * doc/install.texi: Ditto.
33912 * doc/invoke.texi: Ditto.
33914 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33916 * vector-builder.h (vector_builder::m_full_nelts): Change from
33917 unsigned int to poly_uint64.
33918 (vector_builder::full_nelts): Update prototype accordingly.
33919 (vector_builder::new_vector): Likewise.
33920 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
33921 (vector_builder::operator ==): Likewise.
33922 (vector_builder::finalize): Likewise.
33923 * int-vector-builder.h (int_vector_builder::int_vector_builder):
33924 Take the number of elements as a poly_uint64 rather than an
33926 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
33927 from unsigned int to poly_uint64.
33928 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
33929 (vec_perm_indices::new_vector): Likewise.
33930 (vec_perm_indices::length): Likewise.
33931 (vec_perm_indices::nelts_per_input): Likewise.
33932 (vec_perm_indices::input_nelts): Likewise.
33933 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
33934 number of elements per input as a poly_uint64 rather than an
33935 unsigned int. Use the original encoding for variable-length
33936 vectors, rather than clamping each individual element.
33937 For the second and subsequent elements in each pattern,
33938 clamp the step and base before clamping their sum.
33939 (vec_perm_indices::series_p): Handle polynomial element counts.
33940 (vec_perm_indices::all_in_range_p): Likewise.
33941 (vec_perm_indices_to_tree): Likewise.
33942 (vec_perm_indices_to_rtx): Likewise.
33943 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
33944 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
33945 (tree_vector_builder::new_binary_operation): Handle polynomial
33946 element counts. Return false if we need to know the number
33947 of elements at compile time.
33948 * fold-const.c (fold_vec_perm): Punt if the number of elements
33949 isn't known at compile time.
33951 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33953 * vec-perm-indices.h (vec_perm_builder): Change element type
33954 from HOST_WIDE_INT to poly_int64.
33955 (vec_perm_indices::element_type): Update accordingly.
33956 (vec_perm_indices::clamp): Handle polynomial element_types.
33957 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
33958 (vec_perm_indices::all_in_range_p): Likewise.
33959 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
33961 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
33962 polynomial vec_perm_indices element types.
33963 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
33964 * fold-const.c (fold_vec_perm): Likewise.
33965 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
33966 * tree-vect-generic.c (lower_vec_perm): Likewise.
33967 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
33968 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
33969 element type to HOST_WIDE_INT.
33971 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33972 Alan Hayward <alan.hayward@arm.com>
33973 David Sherwood <david.sherwood@arm.com>
33975 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
33976 rather than an int. Use plus_constant.
33977 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
33978 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
33980 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33981 Alan Hayward <alan.hayward@arm.com>
33982 David Sherwood <david.sherwood@arm.com>
33984 * calls.c (emit_call_1, expand_call): Change struct_value_size from
33985 a HOST_WIDE_INT to a poly_int64.
33987 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33988 Alan Hayward <alan.hayward@arm.com>
33989 David Sherwood <david.sherwood@arm.com>
33991 * calls.c (load_register_parameters): Cope with polynomial
33992 mode sizes. Require a constant size for BLKmode parameters
33993 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
33994 forces a parameter to be padded at the lsb end in order to
33995 fill a complete number of words, require the parameter size
33996 to be ordered wrt UNITS_PER_WORD.
33998 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33999 Alan Hayward <alan.hayward@arm.com>
34000 David Sherwood <david.sherwood@arm.com>
34002 * reload1.c (spill_stack_slot_width): Change element type
34003 from unsigned int to poly_uint64_pod.
34004 (alter_reg): Treat mode sizes as polynomial.
34006 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34007 Alan Hayward <alan.hayward@arm.com>
34008 David Sherwood <david.sherwood@arm.com>
34010 * reload.c (complex_word_subreg_p): New function.
34011 (reload_inner_reg_of_subreg, push_reload): Use it.
34013 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34014 Alan Hayward <alan.hayward@arm.com>
34015 David Sherwood <david.sherwood@arm.com>
34017 * lra-constraints.c (process_alt_operands): Reject matched
34018 operands whose sizes aren't ordered.
34019 (match_reload): Refer to this check here.
34021 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34022 Alan Hayward <alan.hayward@arm.com>
34023 David Sherwood <david.sherwood@arm.com>
34025 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
34026 that the mode size is in the set {1, 2, 4, 8, 16}.
34028 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34029 Alan Hayward <alan.hayward@arm.com>
34030 David Sherwood <david.sherwood@arm.com>
34032 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
34033 Use plus_constant instead of gen_rtx_PLUS.
34035 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34036 Alan Hayward <alan.hayward@arm.com>
34037 David Sherwood <david.sherwood@arm.com>
34039 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
34040 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
34041 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
34042 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
34043 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
34044 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
34045 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
34046 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
34047 * config/i386/i386.c (ix86_push_rounding): ...this new function.
34048 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
34050 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
34051 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
34052 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
34053 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
34054 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
34055 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
34056 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
34057 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
34058 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
34059 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
34061 * expr.c (emit_move_resolve_push): Treat the input and result
34062 of PUSH_ROUNDING as a poly_int64.
34063 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
34064 (emit_push_insn): Likewise.
34065 * lra-eliminations.c (mark_not_eliminable): Likewise.
34066 * recog.c (push_operand): Likewise.
34067 * reload1.c (elimination_effects): Likewise.
34068 * rtlanal.c (nonzero_bits1): Likewise.
34069 * calls.c (store_one_arg): Likewise. Require the padding to be
34070 known at compile time.
34072 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34073 Alan Hayward <alan.hayward@arm.com>
34074 David Sherwood <david.sherwood@arm.com>
34076 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
34077 Use plus_constant instead of gen_rtx_PLUS.
34079 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34080 Alan Hayward <alan.hayward@arm.com>
34081 David Sherwood <david.sherwood@arm.com>
34083 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
34084 rather than an int.
34086 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34087 Alan Hayward <alan.hayward@arm.com>
34088 David Sherwood <david.sherwood@arm.com>
34090 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
34091 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
34092 via stack temporaries. Treat the mode size as polynomial too.
34094 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34095 Alan Hayward <alan.hayward@arm.com>
34096 David Sherwood <david.sherwood@arm.com>
34098 * expr.c (expand_expr_real_2): When handling conversions involving
34099 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
34100 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
34101 as a poly_uint64 too.
34103 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34104 Alan Hayward <alan.hayward@arm.com>
34105 David Sherwood <david.sherwood@arm.com>
34107 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
34109 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34110 Alan Hayward <alan.hayward@arm.com>
34111 David Sherwood <david.sherwood@arm.com>
34113 * combine.c (can_change_dest_mode): Handle polynomial
34114 REGMODE_NATURAL_SIZE.
34115 * expmed.c (store_bit_field_1): Likewise.
34116 * expr.c (store_constructor): Likewise.
34117 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
34118 and polynomial REGMODE_NATURAL_SIZE.
34119 (gen_lowpart_common): Likewise.
34120 * reginfo.c (record_subregs_of_mode): Likewise.
34121 * rtlanal.c (read_modify_subreg_p): Likewise.
34123 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34124 Alan Hayward <alan.hayward@arm.com>
34125 David Sherwood <david.sherwood@arm.com>
34127 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
34128 numbers of elements.
34130 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34131 Alan Hayward <alan.hayward@arm.com>
34132 David Sherwood <david.sherwood@arm.com>
34134 * match.pd: Cope with polynomial numbers of vector elements.
34136 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34137 Alan Hayward <alan.hayward@arm.com>
34138 David Sherwood <david.sherwood@arm.com>
34140 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
34141 in a POINTER_PLUS_EXPR.
34143 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34144 Alan Hayward <alan.hayward@arm.com>
34145 David Sherwood <david.sherwood@arm.com>
34147 * omp-simd-clone.c (simd_clone_subparts): New function.
34148 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
34149 (ipa_simd_modify_function_body): Likewise.
34151 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34152 Alan Hayward <alan.hayward@arm.com>
34153 David Sherwood <david.sherwood@arm.com>
34155 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
34156 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
34157 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
34158 (expand_vector_condition, vector_element): Likewise.
34159 (subparts_gt): New function.
34160 (get_compute_type): Use subparts_gt.
34161 (count_type_subparts): Delete.
34162 (expand_vector_operations_1): Use subparts_gt instead of
34163 count_type_subparts.
34165 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34166 Alan Hayward <alan.hayward@arm.com>
34167 David Sherwood <david.sherwood@arm.com>
34169 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
34170 (vect_compile_time_alias): ...this new function. Do the calculation
34171 on poly_ints rather than trees.
34172 (vect_prune_runtime_alias_test_list): Update call accordingly.
34174 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34175 Alan Hayward <alan.hayward@arm.com>
34176 David Sherwood <david.sherwood@arm.com>
34178 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
34180 (vect_schedule_slp_instance): Likewise.
34182 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34183 Alan Hayward <alan.hayward@arm.com>
34184 David Sherwood <david.sherwood@arm.com>
34186 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
34187 constant and extern definitions for variable-length vectors.
34188 (vect_get_constant_vectors): Note that the number of units
34189 is known to be constant.
34191 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34192 Alan Hayward <alan.hayward@arm.com>
34193 David Sherwood <david.sherwood@arm.com>
34195 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
34196 of units as polynomial. Choose between WIDE and NARROW based
34199 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34200 Alan Hayward <alan.hayward@arm.com>
34201 David Sherwood <david.sherwood@arm.com>
34203 * tree-vect-stmts.c (simd_clone_subparts): New function.
34204 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
34206 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34207 Alan Hayward <alan.hayward@arm.com>
34208 David Sherwood <david.sherwood@arm.com>
34210 * tree-vect-stmts.c (vectorizable_call): Treat the number of
34211 vectors as polynomial. Use build_index_vector for
34212 IFN_GOMP_SIMD_LANE.
34214 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34215 Alan Hayward <alan.hayward@arm.com>
34216 David Sherwood <david.sherwood@arm.com>
34218 * tree-vect-stmts.c (get_load_store_type): Treat the number of
34219 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
34220 for variable-length vectors.
34221 (vectorizable_mask_load_store): Treat the number of units as
34222 polynomial, asserting that it is constant if the condition has
34223 already been enforced.
34224 (vectorizable_store, vectorizable_load): Likewise.
34226 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34227 Alan Hayward <alan.hayward@arm.com>
34228 David Sherwood <david.sherwood@arm.com>
34230 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
34231 of units as polynomial. Punt if we can't tell at compile time
34232 which vector contains the final result.
34234 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34235 Alan Hayward <alan.hayward@arm.com>
34236 David Sherwood <david.sherwood@arm.com>
34238 * tree-vect-loop.c (vectorizable_induction): Treat the number
34239 of units as polynomial. Punt on SLP inductions. Use an integer
34240 VEC_SERIES_EXPR for variable-length integer reductions. Use a
34241 cast of such a series for variable-length floating-point
34244 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34245 Alan Hayward <alan.hayward@arm.com>
34246 David Sherwood <david.sherwood@arm.com>
34248 * tree.h (build_index_vector): Declare.
34249 * tree.c (build_index_vector): New function.
34250 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
34251 of units as polynomial, forcibly converting it to a constant if
34252 vectorizable_reduction has already enforced the condition.
34253 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
34254 to create a {1,2,3,...} vector.
34255 (vectorizable_reduction): Treat the number of units as polynomial.
34256 Choose vectype_in based on the largest scalar element size rather
34257 than the smallest number of units. Enforce the restrictions
34260 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34261 Alan Hayward <alan.hayward@arm.com>
34262 David Sherwood <david.sherwood@arm.com>
34264 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
34265 number of units as polynomial.
34267 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34268 Alan Hayward <alan.hayward@arm.com>
34269 David Sherwood <david.sherwood@arm.com>
34271 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
34272 * target.def (autovectorize_vector_sizes): Return the vector sizes
34273 by pointer, using vector_sizes rather than a bitmask.
34274 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
34275 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
34276 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
34278 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
34279 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
34280 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
34281 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
34282 * omp-general.c (omp_max_vf): Likewise.
34283 * omp-low.c (omp_clause_aligned_alignment): Likewise.
34284 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
34285 * tree-vect-loop.c (vect_analyze_loop): Likewise.
34286 * tree-vect-slp.c (vect_slp_bb): Likewise.
34287 * doc/tm.texi: Regenerate.
34288 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
34290 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
34291 the vector size as a poly_uint64 rather than an unsigned int.
34292 (current_vector_size): Change from an unsigned int to a poly_uint64.
34293 (get_vectype_for_scalar_type): Update accordingly.
34294 * tree.h (build_truth_vector_type): Take the size and number of
34295 units as a poly_uint64 rather than an unsigned int.
34296 (build_vector_type): Add a temporary overload that takes
34297 the number of units as a poly_uint64 rather than an unsigned int.
34298 * tree.c (make_vector_type): Likewise.
34299 (build_truth_vector_type): Take the number of units as a poly_uint64
34300 rather than an unsigned int.
34302 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34303 Alan Hayward <alan.hayward@arm.com>
34304 David Sherwood <david.sherwood@arm.com>
34306 * target.def (get_mask_mode): Take the number of units and length
34307 as poly_uint64s rather than unsigned ints.
34308 * targhooks.h (default_get_mask_mode): Update accordingly.
34309 * targhooks.c (default_get_mask_mode): Likewise.
34310 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
34311 * doc/tm.texi: Regenerate.
34313 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34314 Alan Hayward <alan.hayward@arm.com>
34315 David Sherwood <david.sherwood@arm.com>
34317 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
34318 * omp-general.c (omp_max_vf): Likewise.
34319 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
34320 (expand_omp_simd): Handle polynomial safelen.
34321 * omp-low.c (omplow_simd_context): Add a default constructor.
34322 (omplow_simd_context::max_vf): Change from int to poly_uint64.
34323 (lower_rec_simd_input_clauses): Update accordingly.
34324 (lower_rec_input_clauses): Likewise.
34326 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34327 Alan Hayward <alan.hayward@arm.com>
34328 David Sherwood <david.sherwood@arm.com>
34330 * tree-vectorizer.h (vect_nunits_for_cost): New function.
34331 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
34332 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
34333 (vect_analyze_slp_cost): Likewise.
34334 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
34335 (vect_model_load_cost): Likewise.
34337 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34338 Alan Hayward <alan.hayward@arm.com>
34339 David Sherwood <david.sherwood@arm.com>
34341 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
34342 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
34343 from an unsigned int * to a poly_uint64_pod *.
34344 (calculate_unrolling_factor): New function.
34345 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
34347 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34348 Alan Hayward <alan.hayward@arm.com>
34349 David Sherwood <david.sherwood@arm.com>
34351 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
34352 from an unsigned int to a poly_uint64.
34353 (_loop_vec_info::slp_unrolling_factor): Likewise.
34354 (_loop_vec_info::vectorization_factor): Change from an int
34356 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
34357 (vect_get_num_vectors): New function.
34358 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
34359 (vect_get_num_copies): Use vect_get_num_vectors.
34360 (vect_analyze_data_ref_dependences): Change max_vf from an int *
34361 to an unsigned int *.
34362 (vect_analyze_data_refs): Change min_vf from an int * to a
34364 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
34365 than an unsigned HOST_WIDE_INT.
34366 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
34367 (vect_analyze_data_ref_dependence): Change max_vf from an int *
34368 to an unsigned int *.
34369 (vect_analyze_data_ref_dependences): Likewise.
34370 (vect_compute_data_ref_alignment): Handle polynomial vf.
34371 (vect_enhance_data_refs_alignment): Likewise.
34372 (vect_prune_runtime_alias_test_list): Likewise.
34373 (vect_shift_permute_load_chain): Likewise.
34374 (vect_supportable_dr_alignment): Likewise.
34375 (dependence_distance_ge_vf): Take the vectorization factor as a
34376 poly_uint64 rather than an unsigned HOST_WIDE_INT.
34377 (vect_analyze_data_refs): Change min_vf from an int * to a
34379 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
34380 vfm1 as a poly_uint64 rather than an int. Make the same change
34381 for the returned bound_scalar.
34382 (vect_gen_vector_loop_niters): Handle polynomial vf.
34383 (vect_do_peeling): Likewise. Update call to
34384 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
34385 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
34387 * tree-vect-loop.c (vect_determine_vectorization_factor)
34388 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
34389 (vect_get_known_peeling_cost): Likewise.
34390 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
34391 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
34392 (vect_transform_loop): Likewise. Use the lowest possible VF when
34393 updating the upper bounds of the loop.
34394 (vect_min_worthwhile_factor): Make static. Return an unsigned int
34395 rather than an int.
34396 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
34397 polynomial unroll factors.
34398 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
34399 (vect_make_slp_decision): Likewise.
34400 (vect_supported_load_permutation_p): Likewise, and polynomial
34402 (vect_analyze_slp_cost): Handle polynomial vf.
34403 (vect_slp_analyze_node_operations): Likewise.
34404 (vect_slp_analyze_bb_1): Likewise.
34405 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
34406 than an unsigned HOST_WIDE_INT.
34407 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
34408 (vectorizable_load): Handle polynomial vf.
34409 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
34411 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
34413 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34414 Alan Hayward <alan.hayward@arm.com>
34415 David Sherwood <david.sherwood@arm.com>
34417 * match.pd: Handle bit operations involving three constants
34418 and try to fold one pair.
34420 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34422 * tree-vect-loop-manip.c: Include gimple-fold.h.
34423 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
34424 niters_maybe_zero parameters. Handle other cases besides a step of 1.
34425 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
34426 Add a path that uses a step of VF instead of 1, but disable it
34428 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
34429 and niters_no_overflow parameters. Update calls to
34430 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
34431 Create a new SSA name if the latter choses to use a ste other
34432 than zero, and return it via niters_vector_mult_vf_var.
34433 * tree-vect-loop.c (vect_transform_loop): Update calls to
34434 vect_do_peeling, vect_gen_vector_loop_niters and
34435 slpeel_make_loop_iterate_ntimes.
34436 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
34437 (vect_gen_vector_loop_niters): Update declarations after above changes.
34439 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
34441 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
34442 128-bit round to integer instructions.
34443 (ceil<mode>2): Likewise.
34444 (btrunc<mode>2): Likewise.
34445 (round<mode>2): Likewise.
34447 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34449 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
34450 unaligned VSX load/store on P8/P9.
34451 (expand_block_clear): Allow the use of unaligned VSX
34452 load/store on P8/P9.
34454 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34456 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
34458 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
34459 swap associated with both a load and a store.
34461 2018-01-02 Andrew Waterman <andrew@sifive.com>
34463 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
34464 * config/riscv/riscv.md (clear_cache): Use it.
34466 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
34468 * web.c: Remove out-of-date comment.
34470 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34472 * expr.c (fixup_args_size_notes): Check that any existing
34473 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
34474 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
34475 (emit_single_push_insn): ...here.
34477 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34479 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
34480 (const_vector_encoded_nelts): New function.
34481 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
34482 (const_vector_int_elt, const_vector_elt): Declare.
34483 * emit-rtl.c (const_vector_int_elt_1): New function.
34484 (const_vector_elt): Likewise.
34485 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
34486 of CONST_VECTOR_ELT.
34488 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34490 * expr.c: Include rtx-vector-builder.h.
34491 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
34492 directly on the tree encoding.
34493 (const_vector_from_tree): Likewise.
34494 * optabs.c: Include rtx-vector-builder.h.
34495 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
34496 sequence of "u" values.
34497 * vec-perm-indices.c: Include rtx-vector-builder.h.
34498 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
34499 directly on the vec_perm_indices encoding.
34501 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34503 * doc/rtl.texi (const_vector): Describe new encoding scheme.
34504 * Makefile.in (OBJS): Add rtx-vector-builder.o.
34505 * rtx-vector-builder.h: New file.
34506 * rtx-vector-builder.c: Likewise.
34507 * rtl.h (rtx_def::u2): Add a const_vector field.
34508 (CONST_VECTOR_NPATTERNS): New macro.
34509 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
34510 (CONST_VECTOR_DUPLICATE_P): Likewise.
34511 (CONST_VECTOR_STEPPED_P): Likewise.
34512 (CONST_VECTOR_ENCODED_ELT): Likewise.
34513 (const_vec_duplicate_p): Check for a duplicated vector encoding.
34514 (unwrap_const_vec_duplicate): Likewise.
34515 (const_vec_series_p): Check for a non-duplicated vector encoding.
34516 Say that the function only returns true for integer vectors.
34517 * emit-rtl.c: Include rtx-vector-builder.h.
34518 (gen_const_vec_duplicate_1): Delete.
34519 (gen_const_vector): Call gen_const_vec_duplicate instead of
34520 gen_const_vec_duplicate_1.
34521 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
34522 (gen_const_vec_duplicate): Use rtx_vector_builder.
34523 (gen_const_vec_series): Likewise.
34524 (gen_rtx_CONST_VECTOR): Likewise.
34525 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
34526 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
34527 Build a new vector rather than modifying a CONST_VECTOR in-place.
34528 (handle_special_swappables): Update call accordingly.
34529 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
34530 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
34531 Build a new vector rather than modifying a CONST_VECTOR in-place.
34532 (handle_special_swappables): Update call accordingly.
34534 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34536 * simplify-rtx.c (simplify_const_binary_operation): Use
34537 CONST_VECTOR_ELT instead of XVECEXP.
34539 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34541 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
34542 the selector elements to be different from the data elements
34543 if the selector is a VECTOR_CST.
34544 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
34545 ssizetype for the selector.
34547 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34549 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
34550 before testing each element individually.
34551 * tree-vect-generic.c (lower_vec_perm): Likewise.
34553 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34555 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
34556 * selftest-run-tests.c (selftest::run_tests): Call it.
34557 * vector-builder.h (vector_builder::operator ==): New function.
34558 (vector_builder::operator !=): Likewise.
34559 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
34560 (vec_perm_indices::all_from_input_p): New function.
34561 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
34562 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
34563 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
34564 instead of reading the VECTOR_CST directly. Detect whether both
34565 vector inputs are the same before constructing the vec_perm_indices,
34566 and update the number of inputs argument accordingly. Use the
34567 utility functions added above. Only construct sel2 if we need to.
34569 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34571 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
34572 the broadcast of the low byte.
34573 (expand_mult_highpart): Use an explicit encoding for the permutes.
34574 * optabs-query.c (can_mult_highpart_p): Likewise.
34575 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
34576 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
34577 (vectorizable_bswap): Likewise.
34578 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
34579 explicit encoding for the power-of-2 permutes.
34580 (vect_permute_store_chain): Likewise.
34581 (vect_grouped_load_supported): Likewise.
34582 (vect_permute_load_chain): Likewise.
34584 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34586 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
34587 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
34588 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
34589 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
34590 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
34591 (vect_gen_perm_mask_any): Likewise.
34593 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34595 * int-vector-builder.h: New file.
34596 * vec-perm-indices.h: Include int-vector-builder.h.
34597 (vec_perm_indices): Redefine as an int_vector_builder.
34598 (auto_vec_perm_indices): Delete.
34599 (vec_perm_builder): Redefine as a stand-alone class.
34600 (vec_perm_indices::vec_perm_indices): New function.
34601 (vec_perm_indices::clamp): Likewise.
34602 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
34603 (vec_perm_indices::new_vector): New function.
34604 (vec_perm_indices::new_expanded_vector): Update for new
34605 vec_perm_indices class.
34606 (vec_perm_indices::rotate_inputs): New function.
34607 (vec_perm_indices::all_in_range_p): Operate directly on the
34608 encoded form, without computing elided elements.
34609 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
34610 encoding. Update for new vec_perm_indices class.
34611 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
34612 the given vec_perm_builder.
34613 (expand_vec_perm_var): Update vec_perm_builder constructor.
34614 (expand_mult_highpart): Use vec_perm_builder instead of
34615 auto_vec_perm_indices.
34616 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
34617 vec_perm_indices instead of auto_vec_perm_indices. Use a single
34618 or double series encoding as appropriate.
34619 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
34620 vec_perm_indices instead of auto_vec_perm_indices.
34621 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
34622 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
34623 (vect_permute_store_chain): Likewise.
34624 (vect_grouped_load_supported): Likewise.
34625 (vect_permute_load_chain): Likewise.
34626 (vect_shift_permute_load_chain): Likewise.
34627 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
34628 (vect_transform_slp_perm_load): Likewise.
34629 (vect_schedule_slp_instance): Likewise.
34630 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
34631 (vectorizable_mask_load_store): Likewise.
34632 (vectorizable_bswap): Likewise.
34633 (vectorizable_store): Likewise.
34634 (vectorizable_load): Likewise.
34635 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
34636 vec_perm_indices instead of auto_vec_perm_indices. Use
34637 tree_to_vec_perm_builder to read the vector from a tree.
34638 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
34639 vec_perm_builder instead of a vec_perm_indices.
34640 (have_whole_vector_shift): Use vec_perm_builder and
34641 vec_perm_indices instead of auto_vec_perm_indices. Leave the
34642 truncation to calc_vec_perm_mask_for_shift.
34643 (vect_create_epilog_for_reduction): Likewise.
34644 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
34645 from auto_vec_perm_indices to vec_perm_indices.
34646 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
34647 instead of changing individual elements.
34648 (aarch64_vectorize_vec_perm_const): Use new_vector to install
34649 the vector in d.perm.
34650 * config/arm/arm.c (expand_vec_perm_d::perm): Change
34651 from auto_vec_perm_indices to vec_perm_indices.
34652 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
34653 instead of changing individual elements.
34654 (arm_vectorize_vec_perm_const): Use new_vector to install
34655 the vector in d.perm.
34656 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
34657 Update vec_perm_builder constructor.
34658 (rs6000_expand_interleave): Likewise.
34659 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
34660 (rs6000_expand_interleave): Likewise.
34662 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34664 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
34665 to qimode could truncate the indices.
34666 * optabs.c (expand_vec_perm_var): Likewise.
34668 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34670 * Makefile.in (OBJS): Add vec-perm-indices.o.
34671 * vec-perm-indices.h: New file.
34672 * vec-perm-indices.c: Likewise.
34673 * target.h (vec_perm_indices): Replace with a forward class
34675 (auto_vec_perm_indices): Move to vec-perm-indices.h.
34676 * optabs.h: Include vec-perm-indices.h.
34677 (expand_vec_perm): Delete.
34678 (selector_fits_mode_p, expand_vec_perm_var): Declare.
34679 (expand_vec_perm_const): Declare.
34680 * target.def (vec_perm_const_ok): Replace with...
34681 (vec_perm_const): ...this new hook.
34682 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
34683 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
34684 * doc/tm.texi: Regenerate.
34685 * optabs.def (vec_perm_const): Delete.
34686 * doc/md.texi (vec_perm_const): Likewise.
34687 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
34688 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
34689 expand_vec_perm for constant permutation vectors. Assert that
34690 the mode of variable permutation vectors is the integer equivalent
34691 of the mode that is being permuted.
34692 * optabs-query.h (selector_fits_mode_p): Declare.
34693 * optabs-query.c: Include vec-perm-indices.h.
34694 (selector_fits_mode_p): New function.
34695 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
34696 is defined, instead of checking whether the vec_perm_const_optab
34697 exists. Use targetm.vectorize.vec_perm_const instead of
34698 targetm.vectorize.vec_perm_const_ok. Check whether the indices
34699 fit in the vector mode before using a variable permute.
34700 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
34701 vec_perm_indices instead of an rtx.
34702 (expand_vec_perm): Replace with...
34703 (expand_vec_perm_const): ...this new function. Take the selector
34704 as a vec_perm_indices rather than an rtx. Also take the mode of
34705 the selector. Update call to shift_amt_for_vec_perm_mask.
34706 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
34707 Use vec_perm_indices::new_expanded_vector to expand the original
34708 selector into bytes. Check whether the indices fit in the vector
34709 mode before using a variable permute.
34710 (expand_vec_perm_var): Make global.
34711 (expand_mult_highpart): Use expand_vec_perm_const.
34712 * fold-const.c: Includes vec-perm-indices.h.
34713 * tree-ssa-forwprop.c: Likewise.
34714 * tree-vect-data-refs.c: Likewise.
34715 * tree-vect-generic.c: Likewise.
34716 * tree-vect-loop.c: Likewise.
34717 * tree-vect-slp.c: Likewise.
34718 * tree-vect-stmts.c: Likewise.
34719 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
34721 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
34722 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
34723 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
34724 (aarch64_vectorize_vec_perm_const): ...this new function.
34725 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
34726 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34727 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
34728 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
34729 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
34730 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34731 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
34733 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
34734 check for NEON modes.
34735 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
34736 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
34737 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
34738 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
34740 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
34741 the old VEC_PERM_CONST conditions.
34742 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
34743 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
34744 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
34745 (ia64_vectorize_vec_perm_const_ok): Merge into...
34746 (ia64_vectorize_vec_perm_const): ...this new function.
34747 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
34748 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
34749 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
34750 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
34751 * config/mips/mips.c (mips_expand_vec_perm_const)
34752 (mips_vectorize_vec_perm_const_ok): Merge into...
34753 (mips_vectorize_vec_perm_const): ...this new function.
34754 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
34755 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
34756 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
34757 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
34758 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
34759 (rs6000_expand_vec_perm_const): Delete.
34760 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
34762 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34763 (altivec_expand_vec_perm_const_le): Take each operand individually.
34764 Operate on constant selectors rather than rtxes.
34765 (altivec_expand_vec_perm_const): Likewise. Update call to
34766 altivec_expand_vec_perm_const_le.
34767 (rs6000_expand_vec_perm_const): Delete.
34768 (rs6000_vectorize_vec_perm_const_ok): Delete.
34769 (rs6000_vectorize_vec_perm_const): New function.
34770 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
34771 an element count and rtx array.
34772 (rs6000_expand_extract_even): Update call accordingly.
34773 (rs6000_expand_interleave): Likewise.
34774 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
34775 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
34776 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
34777 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
34778 (rs6000_expand_vec_perm_const): Delete.
34779 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
34780 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34781 (altivec_expand_vec_perm_const_le): Take each operand individually.
34782 Operate on constant selectors rather than rtxes.
34783 (altivec_expand_vec_perm_const): Likewise. Update call to
34784 altivec_expand_vec_perm_const_le.
34785 (rs6000_expand_vec_perm_const): Delete.
34786 (rs6000_vectorize_vec_perm_const_ok): Delete.
34787 (rs6000_vectorize_vec_perm_const): New function. Remove stray
34788 reference to the SPE evmerge intructions.
34789 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
34790 an element count and rtx array.
34791 (rs6000_expand_extract_even): Update call accordingly.
34792 (rs6000_expand_interleave): Likewise.
34793 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
34794 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
34796 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34798 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34800 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
34801 vector mode and that that mode matches the mode of the data
34803 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
34804 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
34805 directly using expand_vec_perm_1 when forcing selectors into
34807 (expand_vec_perm_var): New function, split out from expand_vec_perm.
34809 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34811 * optabs-query.h (can_vec_perm_p): Delete.
34812 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
34813 * optabs-query.c (can_vec_perm_p): Split into...
34814 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
34815 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
34816 particular selector is valid.
34817 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
34818 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
34819 (vect_grouped_load_supported): Likewise.
34820 (vect_shift_permute_load_chain): Likewise.
34821 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
34822 (vect_transform_slp_perm_load): Likewise.
34823 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
34824 (vectorizable_bswap): Likewise.
34825 (vect_gen_perm_mask_checked): Likewise.
34826 * fold-const.c (fold_ternary_loc): Likewise. Don't take
34827 implementations of variable permutation vectors into account
34828 when deciding which selector to use.
34829 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
34830 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
34831 with a false third argument.
34832 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
34833 to test whether the constant selector is valid and can_vec_perm_var_p
34834 to test whether a variable selector is valid.
34836 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34838 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
34839 * optabs-query.c (can_vec_perm_p): Likewise.
34840 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
34841 instead of vec_perm_indices.
34842 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
34843 (vect_gen_perm_mask_checked): Likewise,
34844 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
34845 (vect_gen_perm_mask_checked): Likewise,
34847 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34849 * optabs-query.h (qimode_for_vec_perm): Declare.
34850 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
34851 (qimode_for_vec_perm): ...this new function.
34852 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
34854 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34856 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
34857 does not have a conditional at the top.
34859 2018-01-02 Richard Biener <rguenther@suse.de>
34861 * ipa-inline.c (big_speedup_p): Fix expression.
34863 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
34866 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
34869 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
34873 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
34874 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
34875 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
34876 cond_taken_branch_cost 3->4.
34878 2018-01-01 Jakub Jelinek <jakub@redhat.com>
34880 PR tree-optimization/83581
34881 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
34882 TODO_cleanup_cfg if any changes have been made.
34884 PR middle-end/83608
34885 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
34886 convert_modes if target mode has the right side, but different mode
34889 PR middle-end/83609
34890 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
34891 last argument when extracting from CONCAT. If either from_real or
34892 from_imag is NULL, use expansion through memory. If result is not
34893 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
34894 the parts directly to inner mode, if even that fails, use expansion
34897 PR middle-end/83623
34898 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
34899 check for bswap in mode rather than HImode and use that in expand_unop
34902 Copyright (C) 2018 Free Software Foundation, Inc.
34904 Copying and distribution of this file, with or without modification,
34905 are permitted in any medium without royalty provided the copyright
34906 notice and this notice are preserved.